Introduction

Cancer is the second leading cause of death in the United States [1]. Cancer trends appear less favorable among women, compared to men [2]. Breast cancer (BRC) is the leading cancer among women, contributing to most new cancer cases per year in the United States and ranks second in terms of mortality among women [2]. Colorectal cancer (CRC) ranks third in cancer incidence and mortality among women. Both cancers share many similar risk factors, including age, obesity, physical activity status, genetic factors, smoking, and diabetes mellitus [3,4]. Women with diabetes mellitus are at a higher risk of cancer and cancer-related mortality [5,6].

Mammography and colonoscopy are effective evidence-based screening modalities for BRC and CRC, respectively, and have been shown to improve mortality. Despite the proven screening benefits of early detection and improved cancer-related mortality, women with diabetes mellitus are less likely to be screened for BRC/CRC as compared to their non-diabetic counterparts [7–9]. Patients with diabetes mellitus have poorer preventive care and suboptimal adherence to guideline-based screening for BRC/CRC due to co-morbidities [9]. This may, in part, be due to socioeconomic disadvantages, financial barriers, transportation issues, and logistical planning in the outpatient setting [9–14].

Women favor hospitalization time to undergo screening, which may represent a unique opportunity to capture the non-adherent population [15–17]. Therefore, we conducted this study to determine the breast and colorectal cancer screening nonadherence prevalence, associated risk factors, and screening preference among hospitalized women with and without diabetes aged 50-75 years who were cancer-free at baseline.

Methods

Results

Among the 510 hospitalized women participants, 39% had a prior diagnosis of diabetes mellitus. The mean age of participants was 60.5 years (SD =  6.9). Table 1 describes the characteristics of study participants by diabetes status. Hospitalized women with diabetes were more likely to have obesity, ambulation assistive-device dependence (ambulate using a cane, walker, or wheelchair), bedbound status, and were likely on disability or unable to work. Women with diabetes had more medical comorbidities (age adjusted CCI > 3) than their non-diabetic counterparts. However, they were less likely to use alcohol (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of study population.

https://doi.org/10.1371/journal.pone.0319681.t001

Table 2 shows the prevalence of BRC/CRC screening adherence, risk stratification, and patients’ preference for screening mammography and colonoscopies, respectively. Among hospitalized women with diabetes, 28% were nonadherent to screening mammography, and 25% were nonadherent to colorectal cancer screening. 29% never had a screening colonoscopy, whereas 5% never had a screening mammography. There were no statistical differences in the prevalence of nonadherence to either BRC or CRC screening guidelines between hospitalized women with and without diabetes. Additionally, the two groups had no differences in family history or high risk for BRC or CRC. Upon exploring preference for a clinical locale for BRC/CRC screenings, almost half of the hospitalized women with diabetes preferred getting screening mammograms and colonoscopies (46%, p-value 0.62, and 45% p-value 0.23, respectively) if due and offered during hospitalization (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Cancer screening prevalence and preferences of study women while hospitalized.

https://doi.org/10.1371/journal.pone.0319681.t002

We used multivariate logistic regression modeling to evaluate sociodemographic and clinical variables associated with BRC/CRC nonadherence among hospitalized women with diabetes (n = 198). After adjustment of sociodemographic and clinical variables among women with diabetes, the adjusted odds of nonadherence to BRC screening was 3.20 times higher (adjusted OR: 3.20; 95% CI 1.03–9.91) among diabetic hospitalized women who were at high risk for developing CRC as compared to diabetic women at average risk for developing CRC (Table 3). Similarly, adjusted odds of nonadherence to CRC screening were 2.91 times higher (adjusted OR: 2.91; 95% CI 1.15–7.35) in younger women (age less than 60 years) and 2.80 times higher (adjusted OR: 2.80; 95% CI 1.14–6.86) among current smokers or former smokers compared to diabetic women age ≥  60 years, and never smoker respectively (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Predictors of nonadherence to breast and colorectal cancer screening recommendations among hospitalized women with diabetes mellitus only (n = 198).

https://doi.org/10.1371/journal.pone.0319681.t003

Appendix A provides the predictors for BRC and CRC nonadherence among study population without diabetes for reference comparison.

Appendix A

Predictors of nonadherence to breast and colorectal cancer screening recommendations among hospitalized women without diabetes

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

https://doi.org/10.1371/journal.pone.0319681.t004

The post-hospitalization follow-up survey had a low overall response rate of 24% (n = 123). The response rate was slightly higher among women with diabetes at 25% (n = 50) compared to 23% (n = 73) among women without diabetes. Among respondents, 96% (n = 48) of women with diabetes followed up with their primary care provider, compared to 94% (n = 68) of women without diabetes. Regarding screening discussions, 79% (n = 38) of women with diabetes discussed BRC screening and 71% (n = 34) discussed CRC screening with their primary care provider. In contrast, only 57% (n = 83) and 53% (n = 77) of women without diabetes discussed BRC and CRC screening, respectively. Readmission rates were similar, with 54% (n = 27) of women with diabetes reporting readmission compared to 57% (n = 41) of women without diabetes. Post-hospitalization, 62% (n = 31) of women with diabetes received BRC screening, and 38% (n = 19) received CRC screening. Among women without diabetes, 62% (n = 45) underwent BRC screening, and 37% (n = 27) received CRC screening. For women with diabetes who were nonadherent to BRC screening, only 16% (n = 5) received BRC screening post-hospitalization, compared to 84% (n = 26) of diabetic women who were adherent. Similarly, only 11% (n = 2) of women with diabetes who were nonadherent to CRC screening received CRC screening post-hospitalization, compared to 89% (n = 17) of diabetic women who were adherent. None of these differences were statistically significant in the final analysis.

Discussion

This prospective intervention study demonstrates that cancer-free hospitalized women with and without diabetes have alarmingly low rates of self-reported adherence to breast and colorectal cancer at baseline. Almost half of these women report a preference for in-hospital screening, which may be an impactful intervention to improve cancer screening adherence.

Studies have previously shown that women with diabetes have lower BRC and CRC screening rates compared to those without diabetes [8,9]. Our prospective survey-based study negates these findings, although a quarter of women with diabetes reported nonadherence to BRC and CRC cancer screening, (28% and 25% respectively). Lack of appropriate age-related cancer screening in nonadherent women with diabetes poses a population health risk of increased morbidity and mortality, as well as higher healthcare costs.

Our study reports that age < 60 years is associated with lower adherence to BRC/CRC screening, which has been reported previously in African American, Korean-American, and Korean immigrant women regardless of the diabetes status [25]. Younger age may be associated with low adherence rates due to low linkage to outpatient care [26,27]. Smoking (current or past) and nonadherence to BRC screening were also found to be predictors of nonadherence to CRC screening amongst the participants. Smoking status has previously been found to be associated with low likelihood of adherence with breast, endometrial, and colorectal cancer screening in the outpatient setting. Our study found similar results in the inpatient setting for colorectal cancer screening only. Smoking status was not associated with non-adherence to breast cancer screening in our study [28]. Concerted efforts such as smoking cessation programs, mailed/telephonic outreach [29], accessible outpatient appointments [30], and offering inpatient screening interventions for at-risk nonadherent patients may be of value in improving adherence. Although we were unable to find a reasonable explanation for these findings, we suspect that women at high risk of BRC could be more cautious about CRC and perceive cancer risk differently when compared to women at high risk of CRC, primarily due to a lack of awareness of risk. This may be partly due to the lack of a standardized CRC risk stratification tool or the underutilization of comprehensive CRC risk assessment in the outpatient setting (when compared to BRC risk stratification). Also, notably, the societal awareness of BRC screening may be higher due to advertisements, campaigns, etc.

Patients with diabetes often have multiple comorbidities and poor ambulatory capacity, which may make it more difficult for them to navigate outpatient screening interventions in addition to scheduling and transportation challenges. Although our study doesn’t show these factors as predictors of self-reported nonadherence to screening, about half of the women preferred in-hospital screening. This preference may be related to convenience [30]. Post-discharge outpatient adherence was reported to be relatively low for both BRC and CRC, despite more than 90% of women reporting interest in cancer screening in either inpatient or outpatient settings. These findings should prompt healthcare organizations and insurance companies to examine further the benefits and cost-effectiveness of inpatient BRC and CRC screening for select at-risk women with diabetes. Such patient-centered initiatives would help improve screening adherence.

Several limitations of this study should be considered. First, the study was conducted at a single hospital, limiting the findings’ generalizability. Second, the self-reporting methodology has reliability limitations inherent to the survey-based design. Third, our study only included women as part of our study. Breast cancer is the most commonly diagnosed cancer among women [2]. Additionally, female sex is known to be associated with non-compliance with CRC screening as women perceive health and risk of developing CRC differently from men [31,32]. Thus, evaluating prevalence, preferences for screening, and predictors for BRC/CRC nonadherence for women with diabetes should be considered a strength as it highlights gender-specific preferences and barriers to screening. Lastly, the number of patients who responded to the follow-up survey was small, a common problem encountered in many other studies evaluating screening preference and barriers among hospitalized women [32,33]. Nevertheless, the study provides some insight into the pattern of nonadherence to BRC/CRC cancer screening among hospitalized women with diabetes during hospitalization and after discharge from the hospital.

Conclusion

Our study reports that a quarter of cancer-free hospitalized women with and without diabetes are nonadherent to breast and colorectal cancers screening and almost half of these women report a preference for in-hospital screening. Future outcomes research is needed to determine the feasibility and cost-effectiveness of cancer screening in hospital settings.

Acknowledgments

The authors would like to acknowledge the Clinical Excellence Interest Group (CEIG), a Johns Hopkins–wide initiative for scholarly collaboration among clinicians to promote clinical excellence and patient-centered care.