Cervical Cancer Screening among Cambodian-American Women¹

By 1990, census data indicated there were over 1 million Southeast Asians (including ≈150,000 Cambodians) living in the US³ (1). The majority of Cambodians were forced to flee their country because of the political and personal persecution imposed by the Khmer Rouge regime during the mid-1970s and were relocated to North America from refugee camps in Thailand and the Philippines (2, 3). Cambodia is a largely agrarian society, and before the revolutionary period, the majority of Cambodians lived in rural or semirural settings (4). Therefore, immigrants from Cambodia are particularly unfamiliar with Western culture, institutions, and biomedical concepts of early detection (5, 6). There is little information concerning the cancer prevention behavior of Cambodian-American women.

Although the effectiveness of Pap testing has never been evaluated in a randomized controlled trial, observational studies have consistently shown dramatic reductions in mortality rates after the implementation of population-based cervical cancer screening programs (7, 8). Consequently, the American Cancer Society, the National Cancer Institute, and physician specialty organizations uniformly recommend the routine use of Pap testing for all women who have been sexually active or have reached 18 years of age (8, 9, 10). Furthermore, the National Cancer Institute’s year 2000 objectives specify that cervical cancer mortality should be no more than 1.3 per 100,000, that 95% of women should have had at least one Pap smear, and that 85% of women should be receiving regular screenings (11, 12).

The California cancer registry has recently published race-specific data showing that Southeast Asians (Cambodian, Hmong, Laotian, and Vietnamese combined) have markedly elevated invasive cervical cancer incidence and mortality rates. Between 1988 and 1992, age-adjusted incidence rates (per 100,000) were as follows: Southeast Asians, 35.2; Latinas, 17.1; Koreans, 14.7; non-Latina blacks, 12.5; Filipinos, 11.8; Chinese, 8.0; non-Latina whites, 7.5; and Japanese, 5.7. Patterns were similar with respect to the likelihood of dying from cervical cancer: mortality rates varied from ≈2 per 100,000 among non-Latina whites and Japanese to 8.9 per 100,000 among Southeast Asian women (13). In addition, national Surveillance Epidemiology and End Results data for the same time period suggest that cervical cancer is the most commonly occurring malignancy among Vietnamese-American women (incidence rate of 43.0 per 100,000 compared to 37.5 per 100,000 for breast cancer; Ref. 14).

The Pathways to Prevention project recently conducted surveys of five racial/ethnic communities in the San Francisco Bay area. Nearly all of the white (99%) and black (98%) respondents reported at least one Pap smear, compared with 76% of Latina, 67% of Chinese, and 42% of Vietnamese respondents; and the proportions of women reporting Pap testing within the last 3 years ranged from 36% for Vietnamese to 88% for blacks (15). Other population-based studies, conducted in California as well as Massachusetts, have also documented low Pap testing rates among Vietnamese immigrants (16, 17, 18, 19). There is little information about the use of cervical cancer screening by Cambodian-American women. However, Yi (20) surveyed a convenience sample of Cambodian refugees served by community organizations in Houston and found that, although two-thirds (66%) of the respondents had a regular health care provider, only 13% had been screened in the preceding year.

We are currently conducting a randomized controlled trial to evaluate the impact of a neighborhood-based cervical cancer control program targeting Seattle’s Cambodian refugee population. As part of this project, a preintervention survey of Cambodian-American women was completed during late 1997 and early 1998. We used our baseline survey data to examine correlates of ever being screened and, among those with previous Pap testing, the presence of recent screening. Our goal was to provide information about cervical cancer screening barriers and facilitators that could be used to develop effective intervention strategies for Cambodian immigrants.

In Seattle, the Cambodian population is concentrated in the central and south regions of the city (21). In this project, we sought to recruit a representative sample of Cambodian women living in these regions. We developed a sampling frame using multiple sources of data on households located in the target areas. Households were identified from the following sources: electronic listings of all Asian clients of the Seattle Housing Authority and King County Housing Authority, a computer database of motor vehicle licenses from the State of Washington, and two CD-ROM telephone directory databases (SelectPhone and PhoneDisc). From these sources, we selected all households with zip codes corresponding to the target regions and linked the records to a database of Cambodian last names. We then merged records from these sources according to street address to eliminate duplicate entries.

The final sampling frame contained 2145 households, of which 1365 were selected for inclusion in the study. The households were chosen systematically, in that they came from neighborhoods within central and south Seattle that were known to have the highest density of Cambodians, based on 1990 census data. These areas included all residences identified from the two public housing authorities (21).

We publicized the survey through Khmer-language posters distributed in community settings (e.g., Cambodian-owned grocery stores and restaurants), followed by a mailing to each of the study households. The introductory letter, printed in both Khmer and English, was signed by the medical director of Seattle’s International Medicine Clinic and was accompanied by a photograph of the Buddhist Temple at Angkor Wat. All interviews were conducted in respondents’ homes by bilingual, bicultural Cambodian women. Households were offered a calendar featuring traditional Cambodian scenes as a participation incentive and were given the option of answering the survey questions in either Khmer or English.

Women were eligible to participate in the interview if they were 18 years of age or older. When a household included two or more age-eligible women, interviewers asked to speak with the oldest woman. However, if the oldest woman refused or was unavailable, the interviewer asked if a younger female household member would complete the survey. We used this approach rather than a random selection algorithm because attempts to enumerate household members have been shown to reduce response rates in Southeast Asian populations.⁴ Our survey workers made up to five attempts (including at least one daytime, one evening, and one weekend attempt) at contacting each household.

Eyton and Neuwirth (22) have suggested that anthropological methods should routinely be applied during the development of survey instruments for Southeast Asian populations. Our selection of survey questions was guided by an earlier ethnographic study addressing Pap testing barriers and facilitators among Cambodian refugees (23). This qualitative data collection effort allowed us to identify factors relevant to Pap testing among Cambodian women within the context of the diagnostic component of the PRECEDE framework (which was originally taken from Andersen’s model of behavioral factors in health care utilization; Refs. 24, 25, 26). PRECEDE specifies that factors affecting behavior can be broadly classified as predisposing, reinforcing, or enabling. We chose this conceptual framework because, unlike most behavioral models, it assumes that factors affecting health choices are culturally determined and does not specify that the same variables (e.g., perceived susceptibility to disease) are determinants of behavior across communities (25). When appropriate, given the results of our qualitative study, survey items were adapted from the Pathways to Prevention questionnaire, which has been successfully used in several Asian-American populations (15, 17, 27). The survey instrument was developed in English, translated into Khmer, back-translated (to ensure lexical equivalence), and pretested (22).

Women were queried about their age, place of birth in Cambodia (rural versus urban), religion (Buddhist versus other), marital status, educational level, and housing type (owned, rented, or government-subsidized). They also specified how many years they had lived in the US. Women were asked whether they had ever had a Pap test and, if so, when they had their last one. They could respond in number of months, if they had received the test 1 year or less before the survey, or in number of years, if they had received the test more than 1 year before the survey. Because there is a lack of consensus about the need for Pap testing among women without uteri, we also asked each woman if she had a history of hysterectomy (8, 10).

The qualitative study indicated that women’s preventive and traditional (versus biomedical) orientations as well as their Pap testing beliefs would be important predisposing factors. Therefore, each respondent was asked whether she believed the following: illness is a matter of karma, illness is a matter of fate, women should have regular checkups, some diseases are caused by wind illnesses, coin rubbing is the best treatment for some diseases, and some American medicines are too strong for Cambodian people. Additional predisposing factors were whether the survey participants thought Pap smears are necessary for women who are not sexually active, whether they thought Pap testing is necessary for postmenopausal women, and whether they believed that cervical cancer screening can help women live longer.

Enabling factors included past medical history items as well as difficulties accessing health care. Specifically, each survey respondent was asked whether she had ever received prenatal or family planning services in the US and whether she had health insurance coverage. Because a shortage of female doctors was considered a barrier to Pap testing by many of our qualitative interview participants, we asked women with a regular physician to specify his or her sex. Participants were also asked if their access to health care was limited by problems with transportation, problems finding a medical interpreter, and concern about costs. Finally, we considered three potentially reinforcing factors identified by our ethnographic work: previous recommendations for cervical cancer screening by a physician, family member, or friend.

Many Cambodian-Americans have little education (1). Therefore, we made the response options for the PRECEDE items as simple as possible. Specifically, with a few exceptions (e.g., sex of a woman’s physician), the response options were as follows: yes, no, and don’t know/not sure.

We compared the characteristics of women who reported at least one prior Pap smear and women who had never been screened. In a second analysis, we compared women who had and had not received recent cervical cancer screening. For this analysis, women were classified as having been recently screened if they reported their last Pap test was 1 year before the survey or less; those who reported they had never received a Pap test or that their last smear was 2 or more years before the survey were classified as not having been recently screened. Answers to PRECEDE items with response option of yes, no, and don’t know/not sure were dichotomized into yes versus other.

The χ² test and, when necessary, Fisher’s exact test were used to assess statistical significance in bivariate comparisons (28). We used unconditional logistic regression models to summarize concisely the joint effects of sociodemographic variables and PRECEDE factors on cervical cancer screening participation (29). As a tool to build a summary model relevant to intervention planning, we used a forward variable selection method; that is, we entered the most important variables (in terms of deviance change) sequentially into our models until no other variable changed the deviance significantly (30). The final models were scrutinized for scientific interpretability.

Four hundred thirteen women completed the questionnaire. The total estimated household response rate was 72%, and 89% of the households that were reachable and eligible agreed to participate. Twenty-one % of the participating households included more than one age-eligible respondent. The oldest women completed the survey in 79% of the households with two or more eligible women. Six women who reported a personal history of invasive cervical cancer were excluded from the analysis. One further participant was excluded because she did not answer the Pap testing history questions, leaving 406 women. Table 1 gives detailed information concerning the characteristics of the study group.

Three hundred nine (76%) of the 406 women in our study sample reported they had received Pap testing on at least one occasion, and 190 (47%) had been screened recently.

We found significant differences in our bivariate comparisons between “ever” and “never” screened women for age, marital status, housing type, years in the US, and age at immigration (Table 1). Age, marital status, educational level, and age at immigration were all associated with recent screening. The bivariate analyses also showed associations between the following beliefs and Pap testing participation: karma, fate, Pap testing for sexually inactive and post-menopausal women, and prolonging life. Respondents who believed women should get regular checkups were also more likely to have received recent cervical cancer screening. All three of the reinforcing factors considered in this study differed between ever and never screened respondents as well as between women who had and had not received a recent Pap test. Enabling factors associated with Pap testing use included prenatal care and family planning services in the US, having a female provider, and problems finding a medical interpreter. Similar cervical cancer screening participation rates were observed among women with and without health insurance coverage.

Our logistic regression results are summarized in Table 2. The following variables were independently associated with a history of at least one Pap smear: age, years since immigration, beliefs about karma and Pap testing for postmenopausal women, prenatal care in the US, and physician recommendation. Six variables were predictive of recent screening: age, beliefs about regular checkups, cervical cancer screening for sexually inactive women, the prolongation of life, having a female doctor, and a previous physician recommendation for Pap testing.

Pap testing participation rates among Cambodian women in our Seattle sample were higher than those reported from an earlier survey conducted in Texas (only 26% had ever been screened; Ref. 20). It is of note that women who responded to this Houston survey had lower levels of health insurance than our Seattle participants. Discrepancies between the Seattle and Houston results could be due to differences in the sampling methods (the Texas sample included clients of service organizations) or access to health care for economically disadvantaged families. They may also be partly attributable to secular increases, occurring as the Cambodian-American refugee population becomes more acculturated. The Pap testing rates reported by women who responded to our Seattle survey, although much higher than those reported by Yi (20), still do not approach the year 2000 national goals for cervical cancer screening.

We used the PRECEDE model as a conceptual framework for our survey and measures (25, 26). This allowed us to systematically classify factors identified by our qualitative work and facilitate the application of our results to intervention planning. It is of note that multiple predisposing, reinforcing, and enabling factors were associated with Pap testing use in our bivariate analyses. Furthermore, factors from all three of the PRECEDE constructs were independently associated with cervical cancer screening participation. Specifically, beliefs about karma, regular checkups, the prolongation of life, and the necessity of Pap testing (predisposing); prenatal care in the US and sex of provider (enabling); and physician recommendation (reinforcing) were all correlated with screening behavior in one or both of our multivariate models.

Multiple studies, conducted in diverse populations, have demonstrated the association between physician recommendation and women’s use of cancer screening tests (31, 32, 33, 34, 35). We found that Cambodian women who had ever received a physician recommendation for Pap testing were nearly 5 times more likely to have been screened on at least one occasion than those who had never received such a recommendation. However, it should be noted that this information was obtained by self-report and not chart audit. Our results also showed that women with a female health care provider were more likely to have received a recent screening that women with a male provider. This finding may reflect Cambodian women’s reluctance to accept gynecological care from male physicians. Alternatively, it may reflect a greater commitment to routine Pap testing among female providers.

Eighty-six % of our study group had public or private health insurance coverage. However, in contrast to previous studies addressing screening participation, uninsured women were no less likely to have Pap tests than their insured counterparts (16, 31, 33, 36). Additionally, concern about cost was not associated with Pap testing use among our respondents. It is possible that uninsured women are receiving cervical cancer screening through the Centers for Disease Control and Prevention’s Breast and Cervical Cancer Early Detection Program (which provides no-cost services to low-income and uninsured patients).

Our results suggest that programs aimed at increasing the use of Pap testing by Cambodian immigrants might usefully stress that all women should be regularly screening, even if they are postmenopausal and/or not sexually active. We found that survey participants who believed illness is a matter of karma were significantly less likely to have ever had a Pap test than other women. It is not clear why karma is associated with an aversion to technologies of disease prevention such as Pap testing. The philosophical interpretations of karma in Buddhist thought do not necessarily imply passive defeat; Buddhists make common-sense decisions on a daily basis to avoid injury and prevent health problems. How these decisions are made needs to be understood and adapted to disease prevention efforts. It is possible that (once adequately understood) these ideas could be addressed directly by health education programs. Alternatively, related tenets of Buddhism could be adapted to disease prevention efforts. For example, programs could introduce a shift of focus from past karma expressing itself in the present, to improving future karma through “right action” and “making merit” in the present by caring for the self in the interest of children and grandchildren.

Native American women have high cervical cancer incidence and mortality rates compared to the general US population (37, 38). Additionally, Kottke and Trapp (37) have reported that only one-third of Sioux women complete annual Pap testing. Therefore, it is of note that Dignan et al. (38) found that members of the Cherokee and Lumbee tribes have cancer beliefs that are incongruent with biomedical models of prevention. For example, cancer is considered to be a single deadly disease. Furthermore, women feel that finding cancer early is not useful because the emotional turmoil from knowing one has a fatal disease only adds to physical discomfort (38).

There are several limitations to this study that should be considered. (a) We aimed to survey women who would subsequently be randomized to a neighborhood-based intervention trial. Therefore, we chose to only include areas of Seattle with a high density of Cambodian residents. It is unclear to what extent our findings would be generalizable to other geographic regions or Cambodians who do not live in neighborhoods with a high proportion of Southeast Asians. (b) Self-reports of cervical cancer screening use may be inaccurate due to faulty recall or acquiescence bias (i.e., overreporting or a behavior perceived as desirable; Refs. 18,and 39). However, good correlations have been documented between Pap testing rates derived from surveys and chart audits in other populations (18). (c) It is possible that our survey respondents had a different cervical cancer screening experience than women who were unreachable or refused to participate. Specifically, the Pap smear rates among women who responded to the survey may have been higher than those among nonrespondents.

Our results indicate that the complex challenge in promoting the routine use of Pap testing require the precise identification of a population’s early detection barriers and the tailoring of interventions so that they are culturally appropriate. The findings also suggest that disease prevention programs targeting hard-to-reach groups are likely to be more effective if they are multifaceted and simultaneously address predisposing, reinforcing, and enabling factors. For example, an intervention program targeting Cambodian women might usefully include educational materials addressing cervical cancer screening beliefs, education on how to request Pap tests from a physician, and assistance with interpretation at clinic visits. Finally, this study emphasizes the importance of implementing medical office systems that remind physicians to perform interval Pap testing.

Our project works closely with a coalition of women from Seattle’s Cambodian community. We acknowledge the coalition’s contribution to questionnaire development and survey implementation. Additionally, we thank our lead interviewer, Paularita Seng, and the other Cambodian Survey interviewers for their outstanding work.