Mammogram screening in Chile: Using mixed methods to implement health policy planning at the primary care level

Summary

Breast cancer has the highest incidence of all cancers among women in Chile. In 2005, a national health program progressively introduced free mammography screening for women aged 50 and older; however, three years later the rates of compliance with mammographic screening was only 12% in Santiago, the capital city of Chile. This implementation article combines the findings of two previous studies that applied qualitative and quantitative methods to improve mammography screening in an area of Santiago. Socio-cultural and accessibility factors were identified as barriers and facilitators during the qualitative phase of the study and then applied to the design of a quantitative randomized clinical trial. After six months of intervention, 6% of women in the standard care group, 51.8% in the low intensity intervention group, and 70.1% in the high intensity intervention group had undergone a screening mammogram. This review discusses how the utilization of mixed methods research can contribute to the improvement of the implementation of health policies in local communities.

Introduction

Breast cancer is the leading cause of death from malignancies among Latin American women.^1,2 In Chile, the highest incidence of cancer among women is breast cancer. Estimated incidence rates of breast cancer in Chile (49.4/100,000) are in the highest quartile among those reported for Latin American countries.^1,3 Mortality rates from breast cancer in Chile (13.1/100,000) have experienced a slight increase in the last two decades^4-6 and are similar to those reported in Brazil (14.1/100,000) but higher than those observed in Mexico (10.5/100,000).¹ Survival rates from breast cancer in South America are on average 20% lower than those reported in the United States, Western Europe, or Japan.^7,8 This may be at least partially attributed to a more advanced stage at the time of diagnosis.⁷.⁸

The Chilean National Breast Cancer Screening Program began in 1998 and was based on clinical breast examination (CBE). The program achieved high adherence with annual compliance rates of about 65% for the period 2004 to 2007.⁴ Despite the high level of adherence achieved by the CBE program, the percentage of breast cancer cases diagnosed in early stages remained very low. Between 5% and 8% of breast cancer cases were diagnosed in stage 0 (in situ) in 2006.^4,5 This percentage contrasts with the 21.6% of in situ cancer cases reported in the United States SEER registry for the same year.⁹ The low level of early diagnosis and survival rates associated with the slight increase of mortality rates from breast cancer described above suggests a very limited effect of the CBE program in Chile.

In 2005, the Ministry of Health initiated a National Breast Cancer Screening Program that provided inclusion of free mammography screening for women aged 50 and older and universal coverage of diagnostic and therapeutic procedures for women who presented with breast cancer.¹⁰

From the health policy to the local reality

The Catholic University of Chile was interested in conducting a pilot study of the national health policy that allowed free mammography screening for women. A pilot program that included free mammography screening was developed in a university primary care network in the southeast area of Santiago. The national program allowed free screening every two years for women aged 50 to 70 years old. Prior to this pilot study in 2008, the rate of mammography screening was only 12%. The findings observed in Chile were consistent with international evidence that showed that availability of mammography alone is not sufficient for improving mammography utilization.^13-15 High variability of mammography screening rates has been observed in different populations with similar access to mammography screening tests.¹⁶ Even within the same country, screening rates vary significantly across diverse communities. The United States National Health Interview Survey conducted in 2005, for example, showed an average mammography use of 66% by women 40 years or older. However, screening rates varied from 38% to 75%.¹⁷ This high variability among different groups has been attributed to differences in socioeconomic status, race, ethnicity, cultural factors, and regular contact with a general physician.^13,14,17-19

Local Chilean information, as well as the international evidence, emphasized the difficulties of applying an effective health policy such as encouraging regular mammography to age-eligible women. The importance of exploring factors associated with mammography screening practices among Chilean women was needed before conducting a systematic intervention to increase screening. This reality challenged the planning and the implementation processes for improving breast cancer screening and addressed the questions of what had gone wrong in mammography screening in general, and among low socioeconomic women in particular. Thus, the researchers embarked on a dual strategy to: 1) understand why women did not comply with mammography screening; and 2) encourage women to obtain mammography screening. In this paper, we describe the two processes.

Mixed methods as a research strategy for improving health policy implementation

The questions of compliance and encouragement to receive mammography raised in the Chilean experience required a comprehensive methodological approach to better understand the critical factors involved in the implementation of a breast cancer screening program. Such information was necessary to inform the use of different intervention strategies. These are essential steps before formulating and disseminating a public health policy. Following implementation and dissemination research principles^20,21, our group used a mixed methods research model with a sequential design that combined a qualitative phase with a quantitative phase to better understand the lack of compliance with mammography screening.²² The qualitative phase was designed to obtain “deep” information to better understand the barriers and facilitators perceived by Chilean women when it came to obtaining mammography screening. The information gathered in the qualitative phase was applied to design the intervention strategies evaluated in the second phase – the quantitative phase – of the study. In this second phase, a randomized clinical trial was conducted that compared opportunistic screening (usual care) with two intervention strategies of different intensity. The intervention strategies were based on the analysis of the qualitative data.

In previous papers on this topic, we separately examined the qualitative outcomes and the quantitative outcomes^23,24 . Here, we discuss the strength and limitations of the mixed methods research model used as a strategy to improve the implementation of a health policy at the local level. Finally, the authors conclude by examining the importance of using mixed methods in the implementation of a national health policy.

Methods

Mixed Methods

In this review, we discuss “mixed methods” as the strategy used in the research design. Mixed methods strategy blends qualitative and quantitative research to produce a holistic design that uses what is gained from one method (usually qualitative) to develop or inform the other method (usually quantitative). Qualitative research is defined as a design where participants provide their own meaning of the event under investigation. Thus, for this study, we went to the participants (women out of compliance with mammography) and asked them to talk about the barriers that prevented them from having a mammogram. To identify facilitators that led to having a mammogram, we asked compliant women what factors had led them to having a mammogram.

From the qualitative research, we were able to build a quantitative design (randomized controlled trial) that incorporated what the participants believed the meaning of mammography to be as well as the method for testing the efficacy of a program that sought to address both the barriers and facilitators of mammography screening.

Setting

This investigation was carried out in El Castillo Oriente, an underserved area in La Pintana. The municipality of La Pintana is one of the 36 municipalities of the metropolitan area of Santiago, the Chilean capital. About 30% of the population in La Pintana lives below the poverty level. This rate is significantly higher than the 14% of people living in this condition in Santiago.²⁵ The level of education is 33% lower in La Pintana compared to Santiago as a whole. El Castillo Oriente is located within La Pintana and comprises 22,000 people of very low socioeconomic status. This population was registered at a university primary care clinic where women between the ages of 50 and 70 years were able to receive free mammography screening.

Qualitative phase: understanding barriers and facilitators for breast cancer screening

The primary question addressed in the qualitative phase of this study was: “What were the predisposing, enabling and reinforcing factors associated with mammogram screening among women in the study setting at the time of the pilot study?” To address this question, a qualitative study was conducted based on the Predisposing, Enabling and Reinforcing (PRECEDE) framework.²⁶ Details of the methodology used in the study can be found elsewhere.²³ Seven focus groups were conducted with 48 women 50 to 70 years old who have had different experiences with screening practices and diagnosis of breast cancer. The first two groups (G1, G2) included women who had never had a mammogram. The third and fourth groups (G3, G4) included women who had a mammogram during the last two years. The fifth group (G5) was formed by a combination of women who had never had a mammogram and women who had a mammogram in the past two years. The sixth and seventh group (G6, G7), was made up of women who had breast cancer . The qualitative data gathering was approved by the Fred Hutchinson Cancer Research Center Institutional Review Board and by the Ethics Committee of the Pontifica Católica Universidad de Chile in Santiago, Chile. All focus group participants were read an informed consent and signed the form prior to participation. Information was collected using a semi-structured interview schedule, field notes, and field observations, and was audio and video-recorded. The focus groups discussions were transcribed.

The analysis of the qualitative information was conducted following the grounded theory model. Therefore, a sequential process of open, axial and selective coding was used based on the transcriptions. Atlas ti 5.5 software was used for coding and segmenting the data obtained from the interviews.

From qualitative to quantitative methods

For the second quantitative phase, we developed an intervention that incorporated strategies that emerged as important during the qualitative phase. For example, we prepared a brief, tri-fold brochure that addressed shame and secrecy as well as fatalism. These brochures were given in the clinic to the women in the two intervention arms of the RCT (see below). Because women reported not knowing where to obtain a mammogram, we included an order for a mammogram that provided directions to the Mammography Center. Because women thought their physicians needed to recommend mammography, we added a personalized letter from the woman's primary care provider, encouraging her to go to the Mammography Center. Similarly, in the more intensive arm, we sent a promotora (lay health worker) to the woman's home to answer questions the woman may have had about mammography. This additional “cue to action” provided a personal conversation that addressed the barriers uncovered in the qualitative phase of the study.

Quantitative phase: testing different interventions through a randomized clinical trial

Based on the information provided by the qualitative phase of the study, a randomized clinical trial was designed to compare three different interventions for improving mammography screening among women living in La Pintana. A detailed methodology of the study is described elsewhere. ²⁴ Figure 1 presents the study design and follow-up. Basically, a random sample of 500 women 50 to 70 years of age, who had not had a mammogram in the past two years, registered at a community clinic in Santiago and were randomly assigned to one of three intervention approaches. Low intensity intervention included the usual opportunistic advice given to women who contact their primary care provider at the clinic. In addition, the participating women received a mail contact. This mail contact package included a personal letter from the primary care physician, an information brochure and a mammogram order with optional dates for getting the test. High intensity intervention included three components: the first two components were usual opportunistic advice and a mail contact similar to the one described above. The third component was a telephone contact for those women who had not made an appointment for a mammogram after six weeks from receipt of the mail delivery. In case a telephone was not available or the women did not make an appointment for a mammogram in four extra weeks from the previous contact, the women received an in-home visit from a lay health educator. The content of the messages delivered to the participants in both the information booklet and personal contact (telephone or home visit) followed the PRECEDE model that was used to explore barriers and facilitators in a previous phase of the investigation. Basically, predisposing factors such as fatalism related to cancer diagnosis, enabling factors such as clear information about the procedure, and reinforcing factors such as ways to get timely feedback about the results of the test were included in the messages delivered to participants. Standard care intervention included the usual opportunistic advice for women who contacted their primary care physician or midwife. During the opportunistic contact, the health care provider had the option to inform and order a free mammogram for the women who sought care at the clinic.

Figure 1.

Study design and follow-up of randomized clinical trial comparing three strategies for improving mammogram screening in La Pintana, Chile.

The main outcome of the study was compliance with mammography screening. Screening was measured by self-report at baseline and six months after randomization and also through electronic records from the Mammography Center and from the clinic. The sample size of the study was defined based on a point difference (delta) in the screening rate of at least 10% for each intervention arm compared to standard care, a power (beta) of 0.8 and an alpha level of 0.05. Since participants were interviewed at two time points, we estimated an inter-person correlation (P value) of 0.7.

The quantitative phase of the study was also reviewed and approved by the Institutional Review Board at the Fred Hutchinson Cancer Research Center in Seattle and the Ethics Committee at the School of Medicine at Pontificia Universidad Católica de Chile in Santiago.

Results

Qualitative phase: Predisposing, enabling and reinforcing factors for mammogram screening practices

Participants of the focus groups were of low socioeconomic status and had an average of 7.8 years of schooling. Most of them (68%) were married, had, on average, two children and worked as housewives. The average time of each focus group was 1.4 hours (range: 1.2 to 1.8 hours). The total time of content recorded was 10.8 hours (651 minutes).

Figure 2 presents the seven categories from the three PRECEDE dimensions that were identified in the analysis. In the predisposing dimension of the PRECEDE model, socio-cultural factors, knowledge/beliefs and attitudes appeared as the main categories. Fear, shame, secrecy, fatalism, and confidence in the breast self examination (BSE) as the best screening method were the main themes that emerged in this dimension of the model. The mammogram was considered a diagnostic test, i.e. useful only when something was detected in the BSE. As was stated by two informants that had never had a mammogram:

“I don't think I need the test, I've never found anything wrong, thank God” (I6/G2)

“That is why one doesn't go and get it, I will not go to find that I have something wrong, better to live like this’” (I.3/G2)

Figure 2.

Main factors associated with mammogram screening according to the PRECEDE model among women in La Pintana, Chile

Among the enabling factors two categories were analyzed: resources and barriers. Confidence in medical staff, lack of practical information such as where the women could go to receive a mammogram and a high level of bureaucracy and mistreatment were the main themes identified in these categories. The importance of medical advice was highlighted as a key enabling factor by one woman who had breast cancer:

‘Finally, you follow the advice of your physician, the doctor has the final word, he writes the order. When I got cancer, I recommended everybody to have the mammogram, to just go and get the test, but only some of them heard me and all of them went to the doctor first. You don't hear family or friends so much in this”. (I4/G6)

Feedback and reward were the two categories analyzed in the reinforcing dimension of the PRECEDE model. To have timely information of the results, to sense relief, and feel good about herself after getting the test were the main themes that emerged in this category. One participant from the “non- adherent” group emphasized the importance of getting timely feedback for her to be willing to get a mammogram:

“‘I have had the experience of getting tests and then not receiving the results for a long time. I think the same might happen with mammography. If I knew I would get the results soon I would consider getting one.” (I7/G1)

The qualitative phase of the study showed that there was a combination of predisposing, enabling and reinforcing factors that facilitate and/or prevent women from getting mammography screening. Fear, shame and secrecy about breast care were associated with fatalism about finding breast cancer earlier and a perception that the mammogram was mainly a diagnostic test, useful only when something abnormal was found in the BSE.

Confidence in the primary care physician, clear information and timely feedback appeared to be important facilitators of mammography screening.

Quantitative phase: Effective strategies for breast cancer screening

A detailed description and analyses of the quantitative results of this study can be found elsewhere.²⁶ In general; participants were of low socioeconomic status and had a low education level. The majority of them (56.2%) reported an income of less that Ch$ 100,000 (US$ 200) a month and only 22.4% of them completed more than 8 years of education (i.e. high school or university level education). No significant differences between groups were observed in the demographic characteristics of the women.

Six months after randomization, 92.4% (462/500) of the women completed the final survey. The percentage of participants lost to follow-up was similar across the three groups: 6.6% in the control group, 7.8% in low intensity group and 8.4% in the high intensity group. Figure 1 presents the number of women who received the intervention in each group. Brief advice about having a mammogram was delivered by a primary care provider to 15% (73/500) of participants. This intervention had a very similar distribution across groups. In addition to the brief advice, all women in the low intensity group received the mail intervention. In the high intensity group, women received the mail intervention and those who did not get a mammogram were contacted by telephone, and subsequently (if they still did not get a mammogram) received a home visit (outreach intervention). Fifty percent (83/167) of the women in this group were contacted by telephone and/or received a home visit. Forty-three percent of the women (72/167) received both interventions.

Table 1 shows the mammography screening rates after six months in the three groups. There was a significant increase in mammography screening in the low and high intervention arms of the study compared to the control arm. Percentage of screening, according to the electronic registry of mammograms performed during the period of the study, shows a significant increase from 6% (167) in the control arm, to 51.8% (86/166) in the low level intervention arm, and 70.1% (117/167) in the high intensity intervention arm.

Table 1.

Mammogram screening rates by intervention group

Intervention Type
Mammogram Screening	Opportunistic (n=167) N (%)	Low Intensity (n=166) N (%)	High Intensity (n=167) N (%)	Significance
Self-reported mammogram in past 6 months	12/156 (7.7)	79/153 (51.6)	103/153 (67.3)	a,b,c
Electronic record reported mammogram in past 6 months	10/167 (6.0)	86/166 (51.8)	117/167 (70.1)	a,b,c

^aP value for low intensity versus. opportunistic <0.05

^bP value for high intensity versus. opportunistic <0.05

^cP value for low intensity versus high intensity <0.05

Kappa for self-report and electronic record = 0.84

Discussion

The results of the controlled trial showed that an intervention based on qualitative information that addresses specific predisposing, enabling, and reinforcing factors can produce significant improvements in mammography screening practices. Enabling factors related to accessibility, were the most critical ones affected by the intervention. Predisposing factors, especially those related to secrecy, shame and fear were also affected especially when comparing the standard care with the high intensity group (data not shown). The trial also showed that a more intense intervention was associated with a higher effect in mammogram screening practices

The review of these studies shows how applied research that combines qualitative and quantitative methods in a mixed model can produce relevant information that can contribute to the effective application of health policies at a local level. Prior to the study, a low percentage of Chilean women were taking advantage of free mammograms. By the end of the intervention, the rate had improved significantly for both a low level intensity intervention and a highly intense intervention.

The Chilean experience demonstrates an important gap between the design of health policy and its’ effect at the local level. This gap has been widely described in the literature. Haynes et al²⁷ highlighted the existing gap between knowledge and action for health analyzed by the Cochrane Group for Effective Practice and Organization for Care. The Cochrane review emphasized the particular barriers and challenges to implementing health policies in low and middle income countries, such as Chile, due to the relative lack of regulations and quality assurance systems at the local level. Glasgow et al²¹ analyzed the large gap between the results of research studies related to efficacious cancer screening programs and what the programs actually delivered in practice. He presented the key lessons learned regarding dissemination of effective cancer screening interventions.

The Chilean experience, showed that even though the health authorities recognized the limitations of a breast cancer screening strategy such as CBE, and moved to a next step of mammography screening, the local health systems did not respond to the new health policy as was expected. The results of the qualitative phase of the study demonstrated the need for using culturally appropriate strategies to better inform women about the importance of mammography screening and the limitations of BSE for preventing advanced breast cancer.

The results of the controlled trial showed that an intervention based on qualitative information that addresses specific predisposing, enabling and reinforcing factors can produce significant improvements in mammography screening practices. Enabling factors related to accessibility were the most critical factors affected by the intervention. Predisposing factors, especially those related to secrecy, shame and fear were also affected especially when comparing the standard care with the high intensity group. The trial also showed that a more intense intervention was associated with a higher effect in mammogram screening practices.

The sequential process of integrating qualitative and quantitative methods and building mixed models, has been extended in health research during the last decade.²⁸ Our research group has applied mixed methods in a variety of health preventative interventions including randomized community trials for cancer prevention^19,29, controlled clinical trials for smoking cessation^30,31, randomized community trials for reducing pesticide exposure among children^32,33, and breast cancer screening.^25,26 In all cases, the studies have used the main components required in mixed models as described by Creswell et al.²² They arose from the need for specific information about the local implementation of a health policy (rationale), used theory driven qualitative and quantitative methods for data collection and analysis, defined a sequential implementation of methods and integrated them either by triangulation or intervention-design. In the Chilean study, our use of mixed methods applied the PRECEDE model as a general framework for collecting qualitative data and the use of an RCT design to test different intervention in a sequential strategy. Integration of both methods followed the “intervention-design” model where qualitative information was used to design the intervention phase of the study.

From this work, we learned about the importance of identifying the barriers and facilitators to screening. On the face of it, we assumed that free mammograms would yield high rates of compliance; however, we found it took a greater understanding of the barriers to mammography to increase the adherence rate. We also learned that a relatively simple intervention – a mailed approach – could significantly increase adherence. Most importantly, we learned that adding a personal touch, via telephone or face-to-face intervention increased the adherence rate even more dramatically.

This investigation showed how the approach of going back from policy to qualitative research and then forward to quantitative designs could illuminate the local reality and raise opportunities for developing effective interventions.

There were several limitations to the study. First, it took place in a relatively new university clinic system where medical records were electronic and a Mammography Clinic was nearby. There was much support for the project from the university, from the local Community Advisory Board, and from the clinic director. It is possible that implementing such interventions may be more difficult in other settings. Another limitation of the study is associated with maintenance of the effect over time. The intervention was relatively brief; it lasted for six months. The results of the study show that this brief intervention produced a significant effect when compared to the standard care that was developed at the clinic for the last three years. However, maintenance of the effect was not measured, and it is uncertain whether the observed changes last for a longer period of time.

A further limitation is the weakness of mixed methods. Because the qualitative work was based on a particular population, there is no guarantee that it will be generalizable to another population. Although one might argue that poor areas in Santiago might be similar, there is always the possibility that this area was unique. Preparing an intervention for a specific group may preclude accurate hypothesis testing in another group. To prevent this, it may be necessary to replicate the qualitative phase in another population.

Conclusion

The methodology used in the Chilean study provides information on “how” to implement a health policy at a local level with a key component of implementation science³⁴ being qualitative research. However, a key next step will be to find effective ways to disseminate the findings of the study to a broader national level. To what extent will a larger number of clinics apply the components of the strategies tested in this study? What would be the best way to disseminate these findings: using an informational approach, a continuing medical education approach or a community-based program approach? Dissemination strategies at a larger level will require a specific framework to measure the level of adherence of different clinics to the program tested in this trial.³⁵ A dissemination trial would be a natural step to advance the health policy application at a national level.