Effectiveness of an intervention to increase Pap test screening among Chinese women in Taiwan

Effectiveness of an Intervention to Increase Pap Test Screening among Chinese Women in Taiwan

---

Su-I Hou, DrPH, RN

Maria E. Fernandez, PHD

Elizabeth Baumler, PhD

Guy S. Parcel, PhD

Abstract

This study assessed the effectiveness of a theory-based direct mail campaign in encouraging non-compliant women, aged 30 and older or younger if married, to obtain a Pap test. Participants were female family members of inpatients admitted to one of the major teaching hospitals in Taiwan during August–September 1999. A total of 424 women were recruited in the randomized intervention trial. Three months following implementation of the intervention, women in the intervention group reported a higher rate of Pap test screening than women in the comparison group (50% versus 32%) (p = 0.002). Women in the intervention group also showed higher perceived pros of a Pap test at follow-up (p = 0.031). Although women in both groups showed an increased knowledge and intention to obtain a Pap, only the intervention group had significantly higher follow-up scores on perceived pros and susceptibility. The results shows that the intervention was effective in increasing Pap test screening among Chinese women within three months. The results also support the use of Intervention Mapping, a systematic program development framework for planning effective interventions.

Keywords: Pap test screening; intervention mapping; program evaluation; Chinese women.

Introduction

Despite recent declines in cervical cancer incidence and mortality observed in developed countries, these indicators remain high in developing countries.¹ One study reported that the incidence and mortality of cervical cancer were higher in Asia than in the US and other parts of the world.² Cervical cancer is the most prevalent cancer among women in Taiwan. About fourty percent of the female deaths in Taiwan were caused by cancer and about 10 percent of these deaths were due to cervical cancer.³

Pap test screening is the best available method for reducing the incidence and mortality of the invasive cervical cancer.^4–5 Despite the proven efficacy of the Pap test, Asian populations still under-utilize Pap screening. The prevalence of Pap test screening among Asian women in the US ranges from forty-six percent to fifty-two percent, as compared to over ninety percent in other groups.^6–7 The few published studies documenting Pap test screening among Chinese women indicate very low levels of utilization. Among women in Taiwan, existing studies report that regular Pap screening rates are between five percent to forty-two percent.^8–10 Another study reported that about forty percent of women in Taipei had never had a Pap test.¹¹

Only a few studies have identified factors influencing Pap screening among Chinese women. These factors include marital status, age, lack of knowledge, low perceived risk, embarrassment, discomfort, and the gender of physicians.^7,12–14 A pilot study was conducted among a sample of women in Taiwan (1999) as part of the formative evaluation activities for the current study. Analyses revealed factors that were significantly associated with Pap screening behavior. These included knowledge, perceived pros and cons of the Pap test, and perceived norms (reported elsewhere). The study found that lower knowledge level, lower perceived pros, higher perceived cons, and lower perceived norms were strongly associated with non-compliance to recommended Pap test guidelines.

Although there are many educational interventions designed to increase Pap test utilization, few have targeted women in Taiwan. Intervention studies designed to increase screening have provided some information about methods and strategies that have been shown to be effective in other populations. For example, several studies have shown increased screening in response to a written invitations.^15–16 However, another study found the invitation letter alone was not sufficient to encourage women to have a Pap test.¹⁷ One study showed that mailed reminders combined with a telephone contact could significantly increase Pap test compliance.¹⁸

In a review article of cervical cancer screening intervention research, the authors suggested that direct mail may be especially effective in promoting screening when personalized letters or combined strategies were used.¹⁹ The effectiveness of such strategies to increase screening compliance in Taiwan had not been studied. Another gap in the literature is the lack of information about how information from behavioral studies examining the determinants of screening behavior translating into actual intervention methods and strategies.

In Taiwan, the National Health Insurance Plan (1996) provides free health care coverage of annual Pap tests for women aged 30 and over. By paying a little co-payment, women younger than 30 are also entitled to this benefit if married. Removing the cost barrier alone, however, does not ensure screening. It is clear that in Taiwan, although all women have access to free or low cost Pap test screening services, many do not utilize them. This paper describes a study to assess the effectiveness of an innovative intervention that addressed psychosocial factors influencing screening among women in Taiwan. The intervention was built on existing theories, empirical evidence, as well as findings from both qualitative and quantitative research with the target population. Details of the intervention development process are reported elsewhere. The primary objective of this intervention trial is to assess the effectiveness of the intervention in encouraging non-compliant women (who had not had a Pap test within the preceding 12 months) to obtain a Pap test screening within 3 months. The secondary objectives are to assess the impact of the intervention on intermediate factors that might influence Pap screening behavior, including perceived pros/cons of the Pap test, perceived norms, perceived susceptibility, and knowledge of cervical cancer and early detection.

Methods

Sample Selections

This study was conducted at one of the major teaching hospitals in Taichung, Taiwan. Female family members of inpatients admitted to the hospital during August and September 1999 were approached by program staff. A brief enrollment survey was conducted to determine women’s screening status and eligibility. Women who had not had a Pap test in the previous 12 months and were aged 30 and older (or younger if married), were eligible for the study. Women were not eligible if they had undergone a hysterectomy or had been diagnosed with cervical cancer.

A total of 990 women were approached during the study period. Among these, 21 women (0.02%) had a previous hysterectomy. Among the remaining women, sixty-eight percent (656/969) reported they had not had a Pap test in the previous 12 months. Sixty-five percent (424/656) of these non-compliant women agreed to participate in the study. Those who declined participation (N = 232) said that they were not interested or had low literacy level.

Study Design

This evaluation study used a true control group, pretest-posttest experiment design. Individual random assignment to study arm was conducted by assigning consecutive research identification number to each woman as she agreed to participate in the study. Women with even numbers were assigned to the intervention group, while women with odd numbers were assigned to the comparison group.

A total of 424 women were recruited into the study, with 212 women in each group at baseline. A follow-up survey was mailed to these women three months after recruitment. The overall response rate to the mailed follow-up survey was fifty-eight percent (N = 247). Although about forty percent of the women were lost to follow-up, the drop out rate was the same in both intervention and comparison groups. Additionally, there were no significant differences found in demographics between women who dropped-out and those who completed follow-up.

Intervention

Women in the intervention group received a three-month program utilizing direct mail communication as well as a phone-counseling component. In the first month, they received a personalized welcome letter to the study, an educational brochure with theory and evidence-based messages, fourteen quotes regarding other women’s Pap screening experience, and a screening schedule that provided hours for Pap screening with health hotline numbers. In the second month, they received a personalized screening invitation letter, four role model stories with personal accounts from screening compliant women and women who had survived cervical cancer due to early detection, a knowledge-based fact sheet regarding cervical cancer and Pap test, and an updated screening schedule. In the third

month, these women received a telephone call from a health educator to offer barriers counseling and/or assistance with appointment scheduling. Women in the comparison group received a monthly newsletter with health information in general from the hospital.

Measurement

The measurement of the primary outcomes in the survey included items assessing women’s screening behavior and screening intention in the coming year. Other variables of interests included constructs such as perceived pros/cons, perceived norms, perceived susceptibility, and knowledge related to cervical cancer and Pap test screening. Demographic information was also obtained. The items making up the perceived pros, perceived cons, perceived norms, and susceptibility scales used 5-point Likert scale with response categories that ranged from strongly disagree (1) to strongly agree (5). The mean score calculated for each scale was the sum of the item scores divided by total number of items in that scale. The knowledge scale consisted of a sum of correct responses. A pilot study was conducted to assess the psychometric properties of all scales. For each scale, items with discrimination (corrected-item-total correlation CITC) less than 0.2 were re-evaluated for their appropriateness. The internal consistency reliabilities (Chronbach ) resulting from the pilot study were 0.88 for the pros scale (11 items), 0.68 for the cons scale (9 items), 0.72 for the norms scale (4 items), and 0.70 for knowledge scale (16 items) (reference #10). Comments from the open-ended questions in the pilot survey were used to refine the measurement instrument. The final survey used in the current study consisted of thirteen pros items (Chronbach α = .87), thirteen cons items (Chronbach α = .79), four norms items (Chronbach α = .63), three susceptibility items (Chronbach α = .81), and sixteen knowledge items (Chronbach α = .80).

Data Analysis

Chi-square and t-test were used to assess group equivalency at baseline. There were no statistically significant differences found between groups at baseline.

Chi-square tests were used to assess differences between groups on screening completion rates and to assess changes in screening intention among women in the intervention and comparison groups at follow-up. Paired t-tests were used to compare scale differences between pre-test and post-test within each group. Linear regression, including the pre-test scores and group effect as covariates, was used to compare the scale scores between groups at follow-up.

Results

Sample Characteristics and Group Comparisons at Baseline

Comparisons of the Demographics and Prior Screening Between Intervention and Comparison Groups at Baseline. The mean age of the women in the study was 33.87 (SD = 8.61). Most women in the sample were married (90%). Forty percent of the women worked full time and twenty-eight percent of the women had a college education or higher. Prior screening behavior was similar between these two groups. About forty percent of the sample had never had a Pap test. The overall intention to have a Pap test in the coming year was sixty-three percent at baseline and did not differ significantly between groups. There were no significant differences in age, marital status, employment status, or education between women in the intervention and comparison groups. There were no statistically significant differences between groups at baseline in prior screening or intention to have a Pap test in the coming year.

Scales Comparisons Between Groups at Baseline. Knowledge, perceived pros, perceived cons, perceived norms, and perceived susceptibility were also assessed using the scales described above. There were no significant differences on these scales between women in the intervention and comparison groups at baseline. Women in both groups answered about 76% of the knowledge items correctly at baseline. The overall mean scores on the 5-point Likert scale for the perceived pros and cons were 4.17 (SD = 0.41) and 2.78 (SD = 0.54), respectively. The overall means and standard deviations for the perceived norms and susceptibility scales were 3.92 (0.49) and 2.83 (0.70), respectively (Table 1).

TABLE 1: Summary of Scale Means Between Women in Intervention and Comparison Groups (Baseline)

Independent-samples t-tests were used for the baseline group comparisons.

	Intervention (N=212)		Comparison (N = 212)		All (N = 424)
Scales	Mean	SD	Mean	SD	Mean	SD	P-Value (t-test)
Knowledge	0.77	(0.15)	0.74	(0.19)	0.76	(0.17)	0.103
Pros	4.17	(0.39)	4.17	(0.42)	4.17	(0.41)	0.978
Cons	2.77	(0.53)	2.79	(0.54)	2.78	(0.54)	0.748
Norms	3.90	(0.47)	3.95	(0.50)	3.92	(0.49)	0.320
Susceptibility	2.81	(0.68)	2.84	(0.73)	2.83	(0.70)	0.678

 

TABLE 2: Pap Screening Behavior (all women) and Intentions (pre-contemplators only) Between Intervention and Comparison Groups

Pearson chi-square tests were used to compare the differences between intervention and comparison groups.

	Intervention Group		Comparison Group		P-Value (x2) [between group]
	Pre-Test	Post-Test	Pre-Test	Post-Test
Obtain a Pap in the past 3 months
Yes	0 (0%)	63 (51.2%)	0 (0%)	39 (31.5%)	0.002*
No	212 (100%)	60 (48.8%)	212 (100%)	85 (68.5%)
Total	212 (100%)	123 (100%)	212 (100%)	124 (100%)
Intend of a Pap in the coming year
Yes	0 (0%)	43 (86%)	0 (0%)	38 (92.7%)	0.310
No	83 (100%)	7 (14%)	75 (100%)	3 (7.3%)
Total	83 (100%)	50 (100%)	75 (100%)	41 (100%)

Group Comparisons at Post-Test on Screening Completion Rates, Intentions, and Factors Associated with Screening

The follow-up survey was conducted after three months of program implementation. At follow-up, women in the intervention group reported significantly higher rates of Pap screening completion than women in the comparison group (Table 2). Fifty-one percent of previously noncompliant women in the intervention group reported having had a Pap within three months following recruitment, while only thirty-two percent of the women in the comparison group reported receiving a Pap test (p = 0.002).

In this study, women who reported no intention of a Pap screening at baseline were classified as pre-contemplators, whereas women who reported an intention to obtain a Pap test at baseline were classified as contemplators. Changes in intent to obtain a screening between groups among pre-contemplators and the Pap completion rate between groups among contemplators were examined. Among pre-contemplators, the result showed no significant difference in intention change between women in the intervention and comparison groups at follow up (p = 0.310). However, among contemplators, there was a significant difference in Pap completion rate between groups at follow up. A higher the proportion of the contemplators in the intervention group (62%) reported completing a Pap test at followup than women in the comparison group (38%), p = 0.008.

Linear regression models were used to assess differences in scale scores at post-test between women in the intervention and comparison groups on the means of various scales. The analysis included pre-test scale scores and intervention condition as covariates. After adjusting for the pretest scale scores, there were no significant differences found at post-test between groups on the scales measuring knowledge, perceived norms, or perceived susceptibility. Among women in both groups at follow-up, the overall means for knowledge, perceived norms, and perceived susceptibility scales were 0.81, 3.90, and 2.94, respectively (Table 3).

At follow-up, women in the intervention group perceived higher benefits to Pap test compared to women in the comparison group (p = 0.031). The scale means at post-test was 4.26 for the intervention group and 4.16 for the comparison group. The intervention coefficient was 0.110 (95% CI [0.009, 0.181]) for the pros scale. Women in the intervention group perceived lower cons (scale mean = 2.65) of the Pap test than women in the comparison group (scale mean = 2.75). The difference was not that significant, however, the p-value indicated that the difference approaches significance (p = 0.059).

To further examine the relationships between perceived pros and screening behavior, logistic regression analysis was used. The results indicated that perceived pros could predict women’s screening behavior. Women who perceived higher benefits (pros) of a Pap test at follow-up were more likely to have received a screening (OR = 2.45; 95% CI [1.330, 4.503]; p = 0.004), as compared with women who perceived lower benefits of a Pap test screening.

TABLE 3: Summary of Scale Means Between Women in Intervention and Comparison Groups (Follow-up)

Linear regressions were used for the follow-up group comparisons to adjust the pre-test scale and intervention effects.

	Intervention (N=123)		Comparison (N = 124)		All (N = 247)		Standardized Coefficient β	P-Value Adjusted	95% CI for Coefficients β
Scales	Mean	SD	Mean	SD	Mean	SD
Knowledge	0.82	(0.16)	0.79	(0.17)	0.81	(0.17)	0.055	0.305	(−0.017, 0.054)
Pros	4.26	(0.45)	4.16	(0.42)	4.21	(0.43)	0.110	0.031*	(0.009, 0.181)
Cons	2.65	(0.60)	2.75	(0.53)	2.70	(0.57)	-0.088	0.059	(−0.204, 0.004)
Norms	3.88	(0.47)	3.90	(0.48)	3.90	(0.47)	0.008	0.880	(−0.096, 0.112)
Susceptibility	2.96	(0.59)	2.92	(0.71)	2.94	(0.65)	0.048	0.447	(−0.076, 0.173)

Changes in Scales Scores Within Groups Over Time

We conducted further analysis of scales scores to assess changes over time within each group. Although there were no significant differences between intervention and comparison groups on women’s knowledge, perceived norms, and perceived susceptibility, there were significant differences between pre- and post-test times among women within the intervention group on these variables. There were statistically significant differences between pre-test and post-test on knowledge 0.04 (p = 0.016*), perceived pros 0.09 (p = 0.008*), and perceived susceptibility 0.13 (p = 0.011*) among women in the intervention group. For women in the comparison group, knowledge was the only factor that showed significant changes over time. The mean difference was 0.03 (p = 0.02*) (Table 4).

TABLE 4: Within Group Comparisons of Scales Between Pre-Test and Post-Test Within Group Comparisons

Paired t-tests were used to compare the changes of scale means within intervention and control groups.

	Intervention	P-Value	Control	P-Value
Scales	∆Mean	(Paired t-test)	∆Mean	(Paired t-test)
Knowledge	0.04	0.016*	0.03	0.020*
Pros	0.09	0.008*	−0.01	0.838
Cons	−0.08	0.076	0.03	0.472
Norms	0.00	0.961	−0.03	0.470
Susceptibility	0.13	0.011*	0.08	0.101

Discussion

Although there are few studies documenting screening rates among women in Taiwan, our findings seem to coincide with what has been reported by other authors. About forty percent of the participants in this study had never had a Pap test. A similar rate of ‘never been screened’ (38%) among non-compliant women in Kinmen, Taiwan was also found in Lee and Chou’s study.⁹

The primary purpose of the intervention described was to increase Pap test screening among women who were non-compliant to the recommended screening guidelines. The study finding indicates that the intervention was successful to increase women’s Pap test screening behavior. Screening intention was assessed because of the strong relationship between screening intention and actual behavior. We expect to see differences in intention across groups as well as differences in the actual behavior. Although women in the intervention and comparison group showed similar rates of intention to obtain a Pap tests screening at follow-up, changes of intention between pre-and post-test times are significant in each group. Among women who reported no intention to obtain a Pap in the coming year at baseline survey (pre-contemplators), eighty-six percent of the women in the intervention group and ninety-three percent of the women in the comparison group reported an intention at follow-up. This change is likely due to a testing effect on intention.

In an article where the The Transtheoretical Model was applied to the prevention of cancer, Ruggiero underlined one important construct included in the model: decisional balance, which involves the balance between the pros and cons of changing the behavior.²⁰ Decisional balance is critical, especially in the early stage of a behavior change. This model suggests that an individual must increase perceived pros or benefits of changing a behavior in order to move forward from pre-contemplation or contemplation to become ready to take action.²¹ Since all of the recruited women in the current study were either in the pre-contemplation or contemplation stages of a behavior change, the intervention focused on changing the perceived pros of obtaining a Pap test.

Results indicate that the intervention had successfully influenced women’s perceived pros of a Pap test as well as screening behavior. Women who received the intervention showed higher perceived pros scores, and reported higher Pap completion rate than women in the comparison group. Furthermore, the perceived pros were found to be a predictor of women’s Pap test screening behavior (OR = 2.45).

For the within group comparison, perceived norms was the only factor that did not change over time for either group. This indicates that perceived norms (women’s perception about what other women do and think) about the Pap test might be more difficult to influence through the mailed educational packages and phone intervention than the other factors measured.

Because the study population was made up of mostly younger women (age mean = 34), we cannot assume that the intervention would be the same level of effectiveness with older women. This study should be interpreted with the understanding that all the women agreed voluntarily

to participate, and that thirty-five percent of the eligible women refused to participate, therefore, self-selection bias may be functioning. Although we may have expected that these women could have higher screening rates than the general population, similar rates of screening were observed. Data showed about sixty-eight percent (656/969) of the women we approached had not had a Pap in the past year. In Lee and Chou’s study conducted among a similar sample size of women in Taiwan (n = 990), their finding revealed a comparable percentage of non-compliant women (72%).

It is possible that some women were concealing their non-attendance for screening by not responding to the follow-up survey.¹⁶ Although forty percent of the women responding to the baseline survey were lost to follow-up at post-test, the results did not show differential drop-out rates between women in two groups. Additionally, there was no evidence of demographic differences (age, marital status, employment status, or education level) between women who dropped out and women who remained in the study. Women who were lost to follow-up did have lower levels of knowledge and higher perceived cons compared with women who completed the follow-up. Still, among the women were lost to follow-up, there were no significant differences found between women in the intervention and comparison groups on any demographics, prior screening behavior, intentions, as well as any factors associated with Pap screening. Post hoc power calculation revealed the follow-up sample provided a power of eighty-nine percent to detect a difference of twenty percent magnitude in the screening completion rate (assuming α = 0.05).

Although there was some evidence of inaccuracy in the self-reported screening behavior in previous research, the inaccuracy was usually in the direction of over-reporting screening.16 The chance that women who reported not being screened but had in fact been screened was minimal. The self-reported negative predictive value estimates were ninety-five to ninety-four percent.^22–23 Therefore, women in this study at baseline were likely to be a true sample of “under-screened” women. Questions about “when and where” women obtained a Pap test during the study period were asked in order to minimize the possibility of the over-reporting at post-test. If it occurred, it was assumed that the over-reporting rate was non-differential in intervention and comparison groups. Thus, the between group comparisons of screening behavior would not be affected.

The question may be raised as to why we observed a difference in pros scale but not the cons scale. The reason for the lack of significance found in the cons scale may be due to a larger variance observed in the responses of the cons scale compared with the pros scale. Although the mean difference between intervention and comparison groups at post-test was the same for these two scales, the post hoc power analysis for the cons scale indicated a power of only twenty-eight percent to detect a statistically significant differences between groups.

The study results suggest that a theory and evidence-based direct mail intervention can increase Pap screening behavior among Chinese women who were non-compliant to recommended screening guidelines. The direct mail intervention is a cost-effective way to change screening behavior and has the advantage of allowing women to read the information at their leisure. It is also relatively easy to disseminate. Future studies should investigate the role of perceived pros/cons on influencing cancer screening behavior, especially for intervention programs that focus on populations in the early stages of a behavior change. Similar strategies with tailored messages to promote screening behavior can also be adopted to reach at-risk populations.

References

1. Herrero R. Epidemiology of cervical cancer [Review]. J Natl Cancer Inst Monogr 1996;21:1–6.
2. Liu MC, Hai A, Huang AT. Cancer epidemiology in the Far-East—contrast with the United States. Oncology 1993;7:99–110.
3. Lai MY, Zhou BS. Effectiveness of screening for cervical cancer: a case-control Study (thesis). Taipei, Taiwan: National Yang-Ming Medical College. 1990.
4. National Institutes of Health Consensus Development Panel. National Institutes of Health consensus development conference statement: Cervical cancer. J Natl Cancer Inst Monogr 1996;21: vii–xix.
5. Centers for Disease Control and Prevention. The national breast and cervical cancer early detection program; At-A-Glance, 1998. US Dept HHS 1998;1–6.
6. Yi JK. Factors associated with cervical cancer screening behavior among Vietnamese women. J Community Health 1994;19:189–200.
7. Pham CT, McPhee SJ. Knowledge, attitudes, and practices of breast and cervical cancer screening among Vietnamese women. J Cancer Educ 1992;7:305–310.
8. Chou P, Chen L. Mass screening for cervical cancer in Taiwan from 1974–1984. Cancer 1984;64: 962–968.
9. Lee TF, Kuo HS, Chen HC, Chen TS, Chou BS. Factors of Pap screening behavior among women in KinMen, Taiwan. Public Health of Republic of China 1997; 16:198–209.
10. Hou S, Fernandez M, Baumler E, Guy S. Parcel. Cervical cancer screening among women in Taiwan—pilot study. Health Care for Women International (under revisions).
11. Wang PD, Lin RS. Sociodemographic factors of pap smear screening in Taiwan. Public Health 1996;110:123–127.
12. Kelly AW, Fores CM, Wollan PC, et al. A program to increase breast and cervical cancer screening for Cambodian women in a mid-western community. Mayo Clin Proc 1996;71:437–44.
13. Seow A, Wong ML, Smith WCS, Lee HP. Beliefs and attitudes as determinants of cervical cancer screening: A community-based study in Singapore. Prev Med 1995;24:134–141.
14. Yi JK. Acculturation and Pap screening practices among college-aged Vietnamese women in the United States. Cancer Nurs 1998;21:335–341.
15. Byles JE, Redman S, Sanson-Fisher RW, Boyle C. Effectiveness of two direct-mail strategies to encourage women to have cervical (Pap) smears. Health Promotion 1995; 10:5–16.
16. Bowman J, Sanson-Fisher RW, Boyle C, Pope S. A randomized controlled trial of strategies to prompt attendance for a Pap smear. J Med Screen 1995; 2:211–218.
17. Buehler SK, Parsons WL. Effectiveness of a call/recall system in improving compliance with cervical cancer screening: a randomized controlled trial. Can Med Assoc J 1997; 157: 521–526.
18. Lantz PM, Stencil D, Lippert MT, Beversdorf S, Jaros L, Remington PL. Breast and cervical cancer screening in a low-income managed care sample: The efficacy of physician letters and phone calls. Am J Public Health 1995;85:834–836.
19. Marcus AC, Crane LA. A review of cervical cancer screening intervention research: implications for public health programs and future research. Prev Med 1998;27:13–31.
20. Ruggiero Laurie. Transtheoretical Model: Applications in the prevention and treatment of cancer. Med Pediatr Oncol Suppl 1998;1:69–74.
21. Prochaska JO. Strong and weak principles for progressing from pre-contemplation to action based on twelve problem behaviors. Health Psychol 1994;13:47–51.
22. Suarez L; Goldman DA, and Weiss NS. Validity of Pap smear and mammogram self-reports in a low-income Hispanic population. Am J Prev Med 1995; 11:94–98.
23. Walter SD; Clarke EA; Hatcher J, and Stitt LW. A comparison of physician and patient reports of Pap smear histories. J Clin Epidemiol 1988; 41:401–410.