June 2014 VOL 5, NO 3
← Back to Issue
Agreement Between Patient Self-Report and an Objective Measure of Treatment Among African American Women with Breast Cancer
Sue P. Heiney, PhD, RN, FAAN
Background: Only a limited amount of research is focused on African American women with breast cancer and treatment adherence (TA). One potential barrier to TA may be a lack of knowledge and understanding of benefits for the therapies that they receive. To further explore this hypothesis, we examined participants’ recall of their receipt of chemotherapy, radiation, or hormone therapy in comparison to cancer registry data among African American women with breast cancer.
Methods: We compared patient self-report in a sample of 108 African American women with breast cancer with cancer registry data. Univariate conditional logistic regression was used to calculate the odds ratio of discordance for chemotherapy, radiation, and hormone therapy.
Results: There was a high degree of agreement for chemotherapy (95.4%). For radiation, agreement was slightly lower at 83%. For hormone therapy, 72% of the patients were concordant and 28% were discordant. For all covariates, only intervention assignment predicted discordance for radiation therapy at an alpha level of 0.05. The most intriguing finding was the level of discordance between treatment reported in the medical record and treatment reported by the patient.
Conclusion: An important area for future investigations is the relationship between patient education, patient understanding, group support, and TA in African American women with breast cancer. Our findings and future work will provide a foundation for the development of interventions to improve TA.
Treatment nonadherence is associated with increased mortality in women with breast cancer.1-11 It may also partially explain significantly higher mortality in African American women with breast cancer than in white women.12-14 Little data are available that describe treatment adherence (TA) in African American women with breast cancer.12,15
A few studies have focused on TA to radiation therapy or parenteral chemotherapy in African American women with breast cancer,16 although most have examined TA to adjunctive hormonal therapy or oral antineoplastic chemotherapy.15,17-23 Results from these studies have estimated that the proportion of nonadherence ranged between 9% and 46% with only 5.6% to 41% representation of African American women with breast cancer.15,17-23 Only 2 studies were identified that examined oncologic patient factors and adherence.22,23 Kahn and colleagues found that involvement in decision-making and support from healthcare providers positively influenced TA.22
One factor that may negatively impact TA of patients is the lack of understanding of their treatment as measured by accurate treatment recall. In a very small French sample, participants reported a lack of accurate knowledge about the side effects associated with tamoxifen, and the importance of support from healthcare providers in managing side effects.23
Patients’ awareness of and understanding about treatment is an important factor in TA. Neither of these studies examined African American women specifically. The purpose of this investigation was to examine the participants’ recall of receiving chemotherapy, radiation, or hormone therapy compared with data from a cancer registry abstracted from medical records in a large sample of African American women in breast cancer.
We were also interested in determining the agreement between patient self-report and tumor registry data to provide beginning evidence of patient knowledge about their treatment in an understudied population.
We conducted a retrospective cohort study, with a secondary data analysis of self-reported treatment data compared with data obtained from the tumor registry. The parent study, Sisters Tell Others and Revive Yourself (STORY), was a randomized trial comparing the effectiveness of a therapeutic group via teleconference of African American women with breast cancer with usual psychosocial care.24
The STORY study was approved by the Institutional Review Board. Patients signed informed consents and research authorizations giving the study team permission to access their medical information. Data were collected from all participants at baseline, 8 weeks, and 16 weeks. Medical and psychosocial data were collected.
Inclusion/exclusion criteria and recruitment have been described elsewhere and included African American women who had been diagnosed with invasive breast cancer in the previous 6 months and opted for lumpectomy treatment.24,25
The study evaluated the effects of the teleconference group intervention on social connection and identified mediators and moderators of the intervention. Eight weekly 90-minute teleconference sessions were conducted with women randomized to the intervention arm. Knowledge, fear, isolation, and fatalism were also evaluated.
The population for this analysis only included a subset of participants in the original STORY study. Although the STORY study recruited patients across the state of South Carolina, we included only those patients whose data were abstracted by the tumor registry of a local hospital system. This was largely due to this hospital having an established in-house registry and the largest proportion of study participants were from this hospital system.
Data for this investigation were obtained from 3 items on the patient treatment information form. At baseline, patients were asked, “Are you taking any cancer treatment now, yes or no?”, and “If yes, what kind of treatment - radiation and/or chemotherapy?” The third item assessed hormone therapy asking, “Please mark any drug you are taking - goserelin acetate, letrozole, toremifene citrate, anastrozole, fulvertrant” (generic name and trade name were both listed). At time points 2 and 3, patients were asked, “Mark any treatment that you are currently taking - radiation, chemotherapy.” Hormone treatment was assessed using the same question as baseline.
Registry treatment data were obtained by linking the participants of the STORY study with breast cancer cases diagnosed at the local hospital system. This was performed using several patient identifiers, including address, birth date, and social security number when available. There were 185 patients with demographics and time-point assessments; however, only 113 cases were abstracted from the hospital system cancer registry that matched our data set. Four patients were excluded because of missing treatment data and 1 patient was not treated for cancer, which left 108 patients for our analysis.
Treatment data captured within the local hospital cancer registry represent a quality-controlled data element for the central cancer registry. All data were obtained from primary sources such as the hospital’s electronic or paper medical record. As the central cancer registry has extensive quality-control measures and protocols to ensure accurate and timely data, we considered this source an objective (“gold-standard”) measure of their treatment.
We had 3 outcome variables: agreements between patient report and medical record data for chemotherapy, radiation, and hormone therapy. The agreements were coded as “concordant” or “discordant” for each of the 3 types of treatments. We calculated whether patients’ self-reported treatment agreed with the hospital system tumor registry data by creating a series of algorithms in SAS version 9.3 software (SAS Institute; Cary, NC).
Patients were classified as concordant to chemotherapy if they had a record in the hospital system tumor registry and they indicated “yes” that they were taking chemotherapy in any of the time-point assessments. They were also classified as concordant to chemotherapy if they had no record in the hospital system tumor registry and they indicated “no” that they were taking chemotherapy in all of the time-point assessments.
Radiation treatment agreements were calculated using the same methods as chemotherapy agreements. Patients were classified as concordant to hormone therapy if they had a record in the hospital system tumor registry and they indicated that they were taking any of the previously reported drugs at any of the time-point assessments. Patients were classified as discordant to chemotherapy if they had a record in the hospital system tumor registry and they indicated “no” that they were taking chemotherapy in all of the time-point assessments or they indicated “yes” on any time-point assessments and had no record in the tumor registry.
Disagreement for radiation and hormone therapy treatments were calculated using the same algorithms as chemotherapy disagreements. We also compared the date obtained from the hospital tumor registry with the time-point assessments to make sure the treatment date from the hospital preceded the time-point assessments. If the hospital date was after the time-point assessments, then we coded the hospital record as having no treatments (at the time that the patient completed the assessment form).
Due to the small sample size in some of the variables, we categorized majorities of the covariates into dichotomous variables. Age at diagnosis was categorized into “<50” and “>50” years. Education was divided into 2 groups: “<high school” and “>high school.” Relational support was categorized as “partnered or married” and “neither partnered nor married.” Income was categorized into “<$30,000” and “>$30,000.” We categorized comorbidity into “0-2” and “3+.” Resource use evaluation, religiousness, fatalism, and cancer knowledge scores were computed using questionnaires. The scale and computation of the score have been reported previously. For resource use evaluation, religiousness, fatalism, and cancer knowledge, a higher summed score indicated a greater level of the relevant factor.
Frequencies of patients’ characteristics were calculated for each of the treatment methods and by agreements. Two-sided Pearson chi-square, t test, and Fisher exact test were computed (where appropriate) to assess for differences between agreements for each treatment type and covariates. Kappa statistics were calculated among concordant and discordant pairs for each type of treatment. Univariate conditional logistic regression was used to calculate the odds ratio of discordance for chemotherapy, radiation, and hormone therapy. Resource use evaluation, religiousness, fatalism, and cancer knowledge were modeled as continuous variables. All analyses were conducted using SAS and an alpha level of 0.05 was used to determine the significance of all tests.
A total of 108 patients were included in this analysis. The mean age of the patients was 56 years (standard deviation, 11; range, 34-81). The majority (84%) of the patients had at least a high school education or above, approximately 79% were partnered or married, and 43% had 3 or more comorbidities. Table 1 (page 23) presents other characteristics by treatment types and agreements (concordant/discordant). We observed that those individuals who did not accurately recall their chemotherapy treatment were more likely to have a higher income than those who accurately recalled their chemotherapy treatment.
Table 2 displays the degree of agreement between self-report and registry data for chemotherapy, radiation, and hormone therapy treatments. Similar to a correlation coefficient, kappa measures the degree of agreement on a scale of 0 to 1 (0 represents no agreement; 1 represents perfect agreement). Because we were interested not only in the degree of agreement, but also the type of recall (patient recalling treatment when there was none documented in the registry versus patient not recalling treatment when documented in the registry), we provided a description for those patients where disagreement was documented. There was a high degree of agreement for chemotherapy (95.4%, ? = 0.90). For radiation, agreement was slightly lower at 83% (? = 0.46). For hormone therapy, 72% of the patients were concordant and 28% were discordant (? = 0.44). To further examine the direction of the discordance, we compared the report from each source (Table 2).
Predictors of discordance (between self-report and the medical record) in univariate logistic regression models appear in Table 3. The intervention assignment was the only statistically significant variable predicting discordance for radiation therapy at the 0.05 alpha level. Patients who were assigned to the usual treatment group were 3.66 times more likely to disagree with their medical record report of radiation treatment compared with patients who were assigned to the intervention arm.
Because there were relatively small numbers available to explore our hypothesis, a more relaxed statistical significance criterion of .10 was used to evaluate potential predictors of disagreement. At the 0.10 alpha level, those who were older than 50 years were more likely to be in agreement with their medical record report of chemotherapy compared with patients younger than 50 years. In addition, a higher evaluation of resources used was a significant predictor for discordance for radiation therapy, and a greater number of comorbidities were a significant predictor for discordance for hormone therapy.
The purpose of this study was to examine the participants’ recall of receiving chemotherapy, radiation, or hormone therapy compared with cancer registry data among a sample of African American women with breast cancer. We sought to determine the agreement between patient report and medical record data as it would provide evidence of patient knowledge about their treatment in an understudied population. Conceptually, we would expect that patient knowledge or perception of treatment would be inherently linked with their TA. We found evidence for discordance between women’s self-report of their cancer treatment and the medical record. More importantly, there was a larger frequency than expected of women who did not self-report a treatment that was found in the medical record.
It is not surprising to us that women’s recall of hormonal therapy may have more discordance than other treatment modalities because oral treatment could be confused with medications for symptom management or comorbid conditions. These findings are similar to data in the literature for other populations of breast cancer survivors.5,9,15,21 We were surprised to observe that more deliberate and obvious treatments such as chemotherapy and radiation therapy also were not recalled by women for whom the hospital registry did report the treatment. Given that the literature has documented significant nonadherence in African American populations for radiation therapy,26 this may provide some indirect insights into African American breast cancer survivor perceptions of their treatment. This provides indirect evidence of the need of intervention to increase patient knowledge of their treatment modality.
One of the limitations of the research was that we were unable to determine whether discordance in the direction of a reported “yes” by a patient and not captured in the hospital registry was because of the patient’s inaccurate recall or a product of the patient choosing to receive treatment at an external facility. However, because of the unique and limited nature of the local environment for cancer care (only 1 major treatment center), the hospital registry was able to capture complete treatment information on virtually all of the diagnosed patients (personal communication). Therefore, there may be a small number of patients who received treatment at another external facility, which was not captured in the hospital registry. This phenomenon would not have any impact on our findings where patients did not report on therapies that they had been documented as receiving. In addition, we most likely did not have sufficient power to observe all possible relationships for predictors of discordance. Nevertheless, this would have no effect on the significant factors that were observed even with our limited power.
Our study also has several strengths. To our knowledge, this is one of the first studies of its kind to examine self-reported cancer treatment among a rural, African American population. Further research into treatment self-report and adherence has significant implications for decreasing breast cancer mortality disparities in African American women. In addition, we were able to conduct our study within a unique environment in which cancer care was limited to a single provider or medical home that reported to the hospital cancer registry. This study also capitalizes on the linkage of 2 separate data sources: self-report and a hospital tumor registry.
Our findings point to further research avenues, which could prove influential in eliminating cancer disparities. Specifically, our results on the influence of age and comorbidity identify specific populations of women who may have inadequate knowledge of their treatment. We were encouraged that the psychosocial intervention from the parent study, which indicated an improvement of social connection among participants, also appeared to significantly improve patient recall/knowledge of their treatment.
We hypothesize that the “story sharing” during the intervention sessions functioned to increase patient understanding of their treatment modality. More research is warranted to investigate the relationship between patient knowledge and TA using electronic medical records as a comparison source for the agreement analyses.
Funding Source: Funding support was provided by the National Cancer Institute, award number R01CA107305, and supported by the Centers for Disease Control and Prevention, Cooperative Agreement Number U48DP001936.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health or the Centers for Disease Control and Prevention.
Author Disclosure Statement: All authors have nothing to disclose.
Corresponding Author: Sue P. Heiney, PhD, RN, FAAN, 1601 Greene Street, College of Nursing, University of South Carolina, Columbia, SC 29208. E-mail: email@example.com.
At the time a portion of this research was conducted, Heiney was employed as Manager, Psychosocial Oncology at Palmetto Health Cancer Centers in Columbia, SC.
1. Hershman DL, Shao T, Kushi LH, et al. Early discontinuation and non-adherence to adjuvant hormonal therapy are associated with increased mortality in women with breast cancer. Breast Cancer Res Treat. 2011;126(2):529-537.
2. Murphy CC, Bartholomew LK, Carpentier MY, Bluethmann SM, Vernon SW. Adherence to adjuvant hormonal therapy among breast cancer survivors in clinical practice: a systematic review. Breast Cancer Res Treat. 2012;134(2): 459-478.
3. Sedjo RL, Devine S. Predictors of non-adherence to aromatase inhibitors among commercially insured women with breast cancer. Breast Cancer Res Treat. 2011;125(1):191-200.
4. Doggrell SA. Adherence to oral endocrine treatments in women with breast cancer: can it be improved? Breast Cancer Res Treat. 2011;129(2):299-308.
5. Verma S, Madarnas Y, Sehdev S, Martin G, Bajcar J. Patient adherence to aromatase inhibitor treatment in the adjuvant setting. Curr Oncol. 2011;18 (Suppl 1):S3-S9.
6. Sheppard VB, Adams IF, Lamdan R, Taylor KL. The role of patient-provider communication for black women making decisions about breast cancer treatment. Psychooncology. 2011;20(12):1309-1316.
7. Andic F, Godette K, O’Regan R, et al. Treatment adherence and outcome in women with inflammatory breast cancer: does race matter? Cancer. 2011; 117(24):5485-5492.
8. Nekhlyudov L, Li L, Ross-Degnan D, Wagner AK. Five-year patterns of adjuvant hormonal therapy use, persistence, and adherence among insured women with early-stage breast cancer. Breast Cancer Res Treat. 2011;130(2): 681-689.
9. Albert US, Zemlin C, Hadji P, et al. The impact of breast care nurses on patients’ satisfaction, understanding of the disease, and adherence to adjuvant endocrine therapy. Breast Care (Basel). 2011;6(3):221-226.
10. Sharma C, Harris L, Haffty BG, Yang Q, Moran MS. Does compliance with radiation therapy differ in African-American patients with early-stage breast cancer? Breast J. 2010;16(2):193-196.
11. Simons S, Ringsdorf S, Braun M, et al. Enhancing adherence to capecitabine chemotherapy by means of multidisciplinary pharmaceutical care. Support Care Cancer. 2010;19(7):1009-1018.
12. Hershman DL, Unger JM, Barlow WE, et al. Treatment quality and outcomes of African American versus white breast cancer patients: retrospective analysis of Southwest Oncology studies S58814/S8897. J Clin Oncol. 2009;27 (13):2157-2162.
13. Albain KS, Unger JM, Crowley JJ, Coltman CA, Hershman DL. Racial disparities in cancer survival among randomized clinical trials patients of the Southwest Oncology Group. J Natl Cancer Inst. 2009;101(14):984-992.
14. Berz JP, Johnston K, Backus B, et al. The influence of black race on treatment and mortality for early-stage breast cancer. Med Care. 2009;47(9):986-992.
15. Hershman DL, Kushi LH, Shao T, et al. Early discontinuation and nonadherence to adjuvant hormonal therapy in a cohort of 8,769 early-stage breast cancer patients. J Clin Oncol. 2010;28(27):4120-4128.
16. Magai C, Consedine NS, Adjei BA, Hershman D, Neugut A. Psychosocial influences on suboptimal adjuvant breast cancer treatment adherence among African American women: implications for education and intervention. Health Educ Behav. 2008;35(6):835-854.
17. Mayer EL, Partridge AH, Harris LN, Gelman RS, Schumer ST, Burstein HJ, Winer EP. Tolerability of and adherence to combination oral therapy with gefitinib and capecitabine in metastatic breast cancer. Breast Cancer Res Treat. 2009;117(3):615-623.
18. Partridge AH, Avorn J, Wang PS, Winer EP. Adherence to therapy with oral antineoplastic agents. J Natl Cancer Inst. 2002;94(9):652-661.
19. Ruddy K, Mayer E, Partridge A. Patient adherence and persistence with oral anticancer treatment. CA Cancer J Clin. 2009;59(1):56-66.
20. Partridge AH, Archer L, Kornblith AB, et al. Adherence and persistence with oral adjuvant chemotherapy in older women with early-stage breast cancer in CALGB 49907: adherence companion study 60104. J Clin Oncol. 2010;28(14):2418-2422.
21. Kimmick G, Anderson R, Camacho F, Bhosle M, Hwang W, Balkrishnan R. Adjuvant hormonal therapy use among insured, low-income women with breast cancer. J Clin Oncol. 2009;27(21):3445-3451.
22. Kahn KL, Schneider EC, Malin JL, Adams JL, Epstein AM. Patient centered experiences in breast cancer: predicting long-term adherence to tamoxifen use. Med Care. 2007;45(5):431-439.
23. Pellegrini I, Sarradon-Eck A, Soussan PB, et al. Women’s perceptions and experience of adjuvant tamoxifen therapy account for their adherence: breast cancer patients’ point of view. Psychooncology. 2010;19(5):472-479.
24. Heiney SP, Underwood SM, Tavakoli A, Adams SA, Wells LM, Bryant LH. Randomized trial of therapeutic group by teleconference: African American women with breast cancer. Cancer. 2012;118(15):3822-3832.
25. Heiney SP, Adams SA, Wells LM, Johnson H, King JM. Participant evaluation of teleconference group for African American women with breast cancer. Cancer Nurs. 2012;35(2):e24-e30.
26. Rizzo M, Lund MJ, Mosunjac M, et al. Characteristics and treatment modalities for African American women diagnosed with stage III breast cancer. Cancer. 2009;115(13):3009-3015.
Wow, those meetings must be sober events.” I’ve gotten this and other similar comments regularly over the past 6½ years since I received my diagnosis of stage III breast cancer [ Read More ]
The following clinical trials are currently recruiting patients with colorectal cancer for inclusion in several investigations. Each trial description includes the NLM Identifier to use as reference with ClinicalTrials.gov. Adjuvant [ Read More ]