HLRP: Health Literacy Research and Practice

Original Research Open Access

Health Literacy Is an Independent Predictor of Cancer Patients' Hospitalizations

Laura A. Cartwright, PhD; Levent Dumenci, PhD; J. Brian Cassel, PhD; Maria D. Thomson, PhD; Robin K. Matsuyama, PhD

Abstract

Background:

Lower levels of health literacy predict higher hospitalization rates. Hospitalization is expensive and many admissions are potentially avoidable. Research examining the relationship between health literacy and health outcomes for cancer patients is limited, and no published studies to date have focused on hospitalization and health literacy in cancer patients.

Objective:

This study examined the ability of health literacy to predict hospitalization of cancer patients.

Methods:

This secondary data analysis investigated the relationship between health literacy and hospitalization rates of adult cancer patients in the first 5 years after their diagnosis. The sample included a diverse selection of cancer diagnoses and all stages of cancer were represented. Health literacy was assessed using the Cancer Health Literacy Test 30 (CHLT-30), a psychometrically sound measure of health literacy validated in a cancer patient population. Health literacy's ability to predict the number of times hospitalized, total days spent hospitalized, and number of 30-day readmissions was examined using multivariate negative binomial multiple regression to control for the outcomes and potentially confounding variables.

Key Results:

When controlling for potentially confounding variables, a negative relationship was found between health literacy and the number of inpatient hospitalizations (β = −0.041, p = .009) and the total number of days spent hospitalized (β = −0.028, p = .023) independently, whereas the relationship between health literacy and the number of 30-day readmissions failed to reach significance (β = −0.002, p = .903).

Conclusions:

Patients with lower health literacy need additional assistance to avoid unplanned hospitalizations. [Health Literacy Research and Practice. 2017;1(4):e153–e162.]

Plain Language Summary:

Results from this study found that as cancer patients' health literacy scores decreased, the number of times hospitalized and the total number of days spent in the hospital increased. This relationship was significant when controlling for diagnosis, stage at diagnosis, receipt of chemotherapy, number of comorbidities, death, education, and race. Cancer patients with low health literacy are at a higher risk of being hospitalized.

Abstract

Background:

Lower levels of health literacy predict higher hospitalization rates. Hospitalization is expensive and many admissions are potentially avoidable. Research examining the relationship between health literacy and health outcomes for cancer patients is limited, and no published studies to date have focused on hospitalization and health literacy in cancer patients.

Objective:

This study examined the ability of health literacy to predict hospitalization of cancer patients.

Methods:

This secondary data analysis investigated the relationship between health literacy and hospitalization rates of adult cancer patients in the first 5 years after their diagnosis. The sample included a diverse selection of cancer diagnoses and all stages of cancer were represented. Health literacy was assessed using the Cancer Health Literacy Test 30 (CHLT-30), a psychometrically sound measure of health literacy validated in a cancer patient population. Health literacy's ability to predict the number of times hospitalized, total days spent hospitalized, and number of 30-day readmissions was examined using multivariate negative binomial multiple regression to control for the outcomes and potentially confounding variables.

Key Results:

When controlling for potentially confounding variables, a negative relationship was found between health literacy and the number of inpatient hospitalizations (β = −0.041, p = .009) and the total number of days spent hospitalized (β = −0.028, p = .023) independently, whereas the relationship between health literacy and the number of 30-day readmissions failed to reach significance (β = −0.002, p = .903).

Conclusions:

Patients with lower health literacy need additional assistance to avoid unplanned hospitalizations. [Health Literacy Research and Practice. 2017;1(4):e153–e162.]

Plain Language Summary:

Results from this study found that as cancer patients' health literacy scores decreased, the number of times hospitalized and the total number of days spent in the hospital increased. This relationship was significant when controlling for diagnosis, stage at diagnosis, receipt of chemotherapy, number of comorbidities, death, education, and race. Cancer patients with low health literacy are at a higher risk of being hospitalized.

Health literacy, “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions” (Ratzan & Parker, 2000), predicts health outcomes in a variety of patient populations, with lower levels of health literacy being associated with poorer health outcomes (Berkman, Sheridan, Donahue, Halpern, & Crotty, 2011; McNaughton et al., 2014; Mitchell, Sadikova, Jack, & Paasche-Orlow, 2012; Omachi, Sarkar, Yelin, Blanc, & Katz, 2013). However, there has been limited research examining the relationship between health literacy and health outcomes in cancer patient populations (Koay, Schofield, & Jefford, 2012). The majority of existing research involving health literacy of cancer patients has focused on screening behaviors (Bennett, Chen, Soroui, & White, 2009; Ojinnaka et al., 2015; Pagan et al., 2012; Scott, Gazmararian, Williams, & Baker, 2002; White, Chen, & Atchison, 2008), cancer knowledge (Davis et al., 2001; Miller, Brownlee, McCoy, & Pignone, 2007; Peterson, Dwyer, Mulvaney, Dietrich, & Rothman, 2007; Peterson et al., 2011; Wilson et al., 2010), and communication or decision-making (Amalraj, Starkweather, Nguyen, & Naeim, 2009; Friedman, Corwin, Dominick, & Rose, 2009; Kim et al., 2001; Seo, Goodman, Politi, Blanchard, & Kaphingst, 2016; Smith, Dixon, Trevena, Nutbeam, & McCaffery, 2009; Sparks & Nussbaum, 2008) as outcomes.

Empirical research suggests that a relationship exists between health literacy and hospitalization rates. Past research found adults presenting to a walk-in clinic with inadequate health literacy were twice as likely to be hospitalized at least once when compared to those with adequate health literacy (Williams, Baker, Parker, & Nurss, 1998). Lower health literacy has been found to predict increased hospital and emergency department use among patients with chronic obstructive pulmonary disease (Omachi et al., 2013), and increased hospitalization rates among heart failure patients (Wu et al., 2013). A study of general medicine patients identified inadequate health literacy as a risk factor for higher 30-day readmission rates (Mitchell et al., 2012).

Although a significant relationship between health literacy and hospitalization rates appears to exist in several patient populations, no studies to date have examined this relationship exclusively in cancer patient populations. Unplanned hospitalizations are expensive and can be distressing for patients. Brooks et al. (2014) found that 19% of hospitalizations were potentially avoidable in a sample of gastrointestinal cancer patients. Cancer patients are at risk of hospitalization due to their disease or treatment side effects. People with lower health literacy may struggle with following complicated medication regimens or identifying symptoms that require medical attention (Berkman et al., 2011; Kalichman et al., 2008; Evangelista et al., 2010). Understanding the relationship between a cancer patient's health literacy and a subsequent hospitalization would inform routine clinical practice and allow for the development of interventions to decrease potentially avoidable hospitalizations.

This study was guided by the Causal Pathways of Health Literacy framework (Paasche-Orlow & Wolf, 2007). This conceptual model identifies three areas of impact—access and utilization of health care, provider-patient interaction, and self-care—as mediators between health literacy and health outcomes. All three areas of impact likely affect hospitalization rates, but without the existence of a direct effect between health literacy and hospitalization they cannot be evaluated. These mediating variables are outside the scope of this study, so the focus was on the direct effects of health literacy and hospitalization. Using a measure of health literacy developed specifically for use among cancer patients, the Cancer Health Literacy Test 30 (CHLT-30) (Dumenci et al. 2014), this retrospective study described here examined whether health literacy predicted hospitalization rates in a patient population comprised exclusively of cancer patients. This is the first published study to examine this relationship for cancer patients.

Methods

Data Sources

This study was completed as part of a larger R01 study, the Cancer Health Literacy Study (CHLS) (Dumenci et al. 2014), to develop a psychometrically sound measure of health literacy, and it was approved by the Virginia Common wealth University Institutional Review Board. The CHLS produced the Cancer Health Literacy Test 30 (CHLT-30), a valid and reliable measure of health literacy with broad content coverage (Dumenci et al., 2014). The CHLT-30 is administered on a touch-screen device with the multiple-choice items being read to participants. Participants in the CHLS were adults with a history of a cancer diagnosis considered healthy enough by their oncologists to participate in a 1 hour interview. Participants in the CHLS were recruited from an academic, National Cancer Institute-designated, safety net, urban cancer center, and were at varying stages along the cancer care continuum. Participants in the CHLS signed informed consent documents granting researchers permission to collect data from their electronic medical records. Recruitment methods and study design for the CHLS have been described previously (Dumenci et al., 2014).

Data regarding CHLS participants' cancer diagnoses, cancer stage at diagnosis, date of diagnoses, dates of hospitalizations, comorbidities, date of death, and cancer treatments received were provided by the Analytic Services office at Virginia Commonwealth University from the Massey Data Analysis System (MDAS). The MDAS integrates 12 internal and external data sources to provide data on patients.

Eligibility Criteria

Given the goal of this data analysis to examine the relationship between health literacy and hospitalization rates, additional eligibility criteria were applied to create a dataset in which all patients shared similar likelihood of experiencing hospitalization. Participants eligible for inclusion were those interviewed for the CHLS during the 5 years after their first cancer diagnosis, had valid data for their medical record number (MRN), were seen at the cancer center at least once, and did not receive a bone marrow transplant (BMT) or stem cell transplant (SCT) during the first 5 years after their diagnosis. A valid MRN was required to link CHLS data to the electronic medical data from the cancer center. Standards of care surrounding BMT and SCT require extensive hospital stays and incidents of hospitalization are expected. Participants who received such treatment represented a very small percentage of the CHLS sample, so those who received either transplant (n = 26) were excluded.

Sample

Of the 1,306 participants in the CHLS, 752 (57.5%) met all inclusion criteria. Demographic and clinical information, including comorbidities and cancer diagnostic information for the final dataset, can be seen in Table 1.

Demographic VariablesDemographic Variables

Table 1:

Demographic Variables

Variables Included in Model Building

Variables in analyses include health literacy, demographic variables, cancer-related variables, and a number of comorbidities for each participant. Also included in the modeling are two variables intended to address limitations in the study: (1) distance from participants home to the cancer center and (2) time period of data analysis. Death was included in model building for participants who died within the time period of analysis.

CHLT-30 score. The ability of health literacy, as measured by the CHLT-30, to predict hospitalization is the primary predictor of interest in these analyses. The CHLT-30 produces continuous health literacy scores ranging from 0 to 30, with a higher score indicating a higher level of health literacy (Dumenci et al., 2014).

Demographics. Education and race are the demographic variables included in models. Education level was accounted for in model building by including a dichotomous variable that categorized participants as having a high school diploma, general education diploma, or less education, or having more than a high school diploma. Race was accounted for by indicating if participants identified as non-Hispanic minority or non-Hispanic White.

Cancer-related covariates. A participant's cancer diagnosis, cancer stage at diagnosis, and receipt of chemotherapy were included in model building. Participants represented a wide variety of cancer diagnoses. Including diagnosis as a categorical variable to indicate each unique diagnosis would have introduced undue complexity to the models. Given that standards of care often require patients with hematologic cancers to have long, planned hospitalizations for treatment, a flag variable was created to indicate if a person had a hematologic cancer or a nonhematologic cancer.

A flag variable indicating if a patient had a stage IV disease at diagnosis was created and included in model building. Stage IV is the final stage of cancer and represents the most severe disease progression. Identifying those with stage IV disease at diagnosis was designed to help account for varying severity of disease across patients.

Given that chemotherapy can affect a patient's likelihood of hospitalization, a flag variable was created to indicate if a person received chemotherapy at any point during the time period of data analysis.

Comorbidities. The number of comorbidities a person was diagnosed with during the time period of analysis was included in model building. Comorbidities included are those identified in the Charlson Comorbidity Index (CCI), a commonly used and validated comorbidity measure (Charlson, Pompei, Ales, & MacKenzie, 1987). The diseases used when calculating the CCI include myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatic disease, peptic ulcer disease, mild liver disease, diabetes without complications, diabetes with complications, cancer, paraplegia or hemiplegia, renal disease, moderate or severe liver disease, metastatic carcinoma, and AIDS/HIV (Quan et al., 2005). Metastatic cancer was included in the comorbidity count as it indicates a change in disease progression and mortality, regardless of what the stage of cancer may have been at diagnosis. Comorbidities were identified using the International Statistical Classification of Diseases and Related Health Problems, 9th Rev. (1979) codes developed by Quan et al. (2005).

Time period of data collection. The focus of data analysis for each participant is the period of 5 years after diagnosis. However, not all participants have a full 5 years of data due to death or timing of interview. The variable time period was calculated to represent the number of years of data available for each participant.

Distance from home to cancer center. How far away a person lived from the cancer center was thought to potentially affect their likelihood of hospitalization at the cancer center versus another hospital. To account for this and to address the limitation of using institutional data that does not document hospitalizations outside of the institution, travel time in minutes from the participants' home zip code to the cancer center was calculated.

Outcome variables. The outcomes of interest in these analyses are hospitalization rates. Hospitalization rates were operationalized as the total number of inpatient hospital admissions, the total number of days spent hospitalized, and the total number of times a participant was readmitted to the hospital within 30 days of discharge during the time period of analysis. All-cause hospitalizations were included to capture participants' experiences during the time period of analysis. The ranges, means, and standard deviations for the outcome variables appear in Table 2. Histograms for each outcome can be seen in Figures 13.

M1 CHLT-30 Multivariate Model Outcomes

Table 2:

M1 CHLT-30 Multivariate Model Outcomes

Total number of hospital admissions.

Figure 1.

Total number of hospital admissions.

Total number of days hospitalized.

Figure 2.

Total number of days hospitalized.

Total number of 30-day readmissions.

Figure 3.

Total number of 30-day readmissions.

Analysis

The outcome variables of interest, total number of times admitted to the hospital, total number of days spent hospitalized, and number of 30-day readmissions are positively skewed count data demonstrating overdispersion (Figure 1), such that the variance and means are not equal (p < .001). Given the characteristics of the outcome variables, multivariate negative binomial multiple regression was used in model testing. Multivariate modeling allows for including more than one outcome variable, so all three hospitalization outcomes are included in each model iteration. Multivariate multiple regressions were completed using MPlus software, version 7.4 (Muthen & Muthen, 2010). The Type-I error rate was set to .05. Model building involved three iterations. The first model (M1) included only the CHLT-30 score and the three outcomes variables. The second model (M2) involved using all the covariates of interest to predict both the CHLT-30 and the outcome variables. The CHLT-30 was a predictor of the outcome variables in this model. The second model informed the structure of the third model (M3), in which the CHLT-30 and the outcome variables were regressed only on significant predictors identified in M2. Given that the CHLT-30 score is the primary predictor of interest, it was regressed on to the hospitalization outcomes in M3 regardless of the significance found in M2. The log of the time period was used in analyses to accommodate negative binomial regression computations done on the log scale (Lawless, 1987).

Results

CHLT-30 scores of this sample ranged from 6 to 30, with a mean of 23.68 (standard deviation [SD] = 5.52). The majority of participants were non-Hispanic White (61%), had beyond a high school education (64.5%), had a nonhematologic cancer (73.7%), and received chemotherapy as treatment for their cancer (63%). Approximately one-half of the participants were women (49.9%). The total number of comorbidities any one participant had ranged from 0 to 9 diagnoses, with a mean of 1.38 diagnoses (SD = 1.49). The number of participants with each individual diagnosis can be seen in Table 1. Time period ranged from 0.31 years to 5 years with a mean of 3.6 years (SD = 1.27). Travel time in minutes ranged from 6 minutes to 299 minutes, with a mean of 34.74 minutes (SD = 30.03).

In the first model, M1, there was a significant relationship between CHLT-30 score and the outcome variables number of admissions (β = −0.074, p < .001) and total number of days hospitalized (β = −0.024, p = .015) seen in Table 2.

Table 3 gives the model parameters for M2, in which all covariates described previously and the CHLT-30 scores are included. The covariates stage IV, race, chemotherapy, hematology flag, deceased, and comorbidity count were found to significantly predict at least one of the hospitalization outcomes. Education, race, and comorbidity count were found to significantly predict CHLT-30 score. The covariates log of time period and travel time did not reach significance in predicting any of the hospitalization outcomes and were therefore not included in M3.

M2 CHLT-30 Multivariate Model Outcomes

Table 3:

M2 CHLT-30 Multivariate Model Outcomes

In the final multivariate multiple negative binomial model (M3), the CHLT-30 score and the covariates chemotherapy, stage IV, hematology flag, deceased, comorbidity count, and race significantly predicted at least one of the outcome variables. CHLT-30 scores were significantly predicted by comorbidity count, race, and education. Parameters for M3 are given in Table 4.

M3 CHLT-30 Multivariate Model Outcomes

Table 4:

M3 CHLT-30 Multivariate Model Outcomes

The direct relationship between CHLT-30 score and number of inpatient hospital admissions (β = −0.041, p = 0.009) and total number of days spent hospitalized (β = −0.028, p = .023) were statistically significant. This model shows that for every 1-point increase in CHLT-30 score, participants experienced a 4% decrease in number of inpatient hospital admissions and a 2.8% decrease in total number of days spent hospitalized, when controlling for covariates and the two other outcomes. Direct effects between CHLT-30 and number of 30-day readmissions (β = −0.002, p = .903) did not reach significance.

In M3, the number of inpatient hospitalization admissions was significantly predicted by receiving chemotherapy (β = 0.630, p < .001), having a hematologic cancer (β = 0.696, p < .001), dying (β = 0.571, p = .001), and number of comorbidities (β = 0.354, p < .001). Total number of days spent hospitalized was significantly predicted by receiving chemotherapy (β = 0.823, p < .001), having a stage 4 cancer (β = −0.365, p = .017), having a hematologic cancer (β = 0.646, p < .001), dying (β = 0.740, p < 0.001), number of comorbidities (β = 0.325, p < .001), and race (β = 0.233, p = .024). The number of 30-day readmissions was significantly predicted by receiving chemotherapy (β = 0.913, p < .001), having a hematologic cancer (β = 1.242, p < .001), dying in the time period (β = 0.601, p = .003), and number of comorbidities (β = 0.310, p < .001).

Discussion

Findings from this study indicate that lower health literacy, as measured by the CHLT-30, is associated with a higher number of inpatient hospitalizations and total number of days spent hospitalized for cancer patients in the first 5 years postdiagnosis. This relationship remains significant after potentially confounding variables are included in the model. The use of multivariate modeling allows us to examine the unique contribution of health literacy score to number of inpatient hospitalizations and total number of days hospitalized by estimating parameters simultaneously. This means even when controlling for the number of times a person is hospitalized, lower CHLT-30 score predicts a higher number of days spent hospitalized. Number of 30-day readmissions did not reach significance in any of the model iterations.

Our study contributes to the mixed findings regarding the relationship between health literacy and hospitalization. A significant inverse relationship between health literacy and the number of inpatient admissions has been found in a number of other patient populations using varying measures of health literacy (Baker, Parker, & Williams, 1998; Omachi et al., 2013; Wu et al., 2013), although Apter et al. (2013) did not find health literacy significantly predicting hospitalization among patients with moderate to severe asthma. Our study did not replicate a past study finding health literacy significantly predicting a 30-day readmission (Mitchell et al., 2012). As is frequently the case in health literacy research, the varying patient populations, types of health literacy measures, and scoring techniques used make comparing findings regarding hospitalization across studies difficult. However, it does seem clear that health literacy plays a role as an independent risk factor for hospitalization for some patient populations.

Cancer diagnosis and treatment is a complicated process for anyone. Patients with lower health literacy may particularly struggle to understand their disease, treatment and medication regimens, and symptom management and identification during an already challenging time. Given that much of cancer treatment is administered in a clinic by a clinician, our findings indicate patients may be experiencing difficulties managing their self-care. Short of not showing up for appointments, individual patients have minimal participation in their chemotherapy, radiation, or surgical treatments. However, cancer patients are potentially responsible for taking medication, performing wound care, maintaining a healthy weight via proper nutrition, and protecting themselves from illness when immunocompromised. All of these self-care tasks are difficult for people with lower health literacy and may result in hospitalization if improperly managed.

Health literacy level was identified as an independent predictor of hospitalization even when controlling for education. Physicians can't assume based on education level that a person has the skills necessary to fully understand the care plan or be able to communicate questions or concerns effectively. Knowing a patient's health literacy level would help care providers identify people who may need additional assistance understanding information and who may struggle in accessing information they need later.

Given that all-cause hospitalizations were included in these analyses, these findings may reflect difficulties people with low health literacy experience navigating the health care system. People with limited health literacy receive less preventive care (Scott et al., 2002) and are more likely to struggle with access to care (Berkman et al., 2011), which could lead to increased hospitalization. Future research should examine the specific causes of hospitalizations to fully understand what is causing those with lower health literacy to be hospitalized longer and more frequently.

An interesting finding in this study was that non-Hispanic White participants had significantly more total days spent in the hospital, but not significantly more hospital admissions, as compared to non-Hispanic minority participants. This finding means that even though non-Hispanic White patients are not hospitalized more often, they do have longer hospital stays. Holding other covariates and outcome variables in the model constant, non-Hispanic White participants are expected to have 26% more days spent in the hospital as compared to non-Hispanic minority participants. Although health disparities research has identified racial differences in cancer treatment recommendations (Li, Malone, & Daling, 2003; Shavers & Brown, 2002) and choices (Hurwitz et al., 2016; Ross et al., 2016), research examining hospital length of stay has found non-White cancer patients have longer stays as compared to their White counterparts (Parsons, Habermann, Stain, Vickers, & Al-Refaie, 2012; Ravi et al., 2015). A potential explanation for our conflicting finding may be due to the heterogeneity of the cancer diagnoses of participants in this study. Although analysis accounted for hematologic cancers, which may have characteristically longer planned hospital stays for treatment, it is possible that racial distribution across the cancer diagnoses could further explain this surprising finding. Future work should be done to examine the potential role diagnosis could play between race and number of days spent in the hospital.

Study Limitations

Participants in the CHLS were people with a history of a cancer diagnosis deemed healthy enough to participate by their oncologist. It is possible that patients who responded well to treatment, were overall healthier, or who had less severe disease were more likely to participate in the CHLS. This could create a sample of people less likely to be hospitalized. Although this is a possibility, the sample of participants in these analyses represented a variety of diagnoses, represented the four stages of cancer roughly equally, and had a number of comorbidities.

These analyses were retrospective in nature and were limited in that only data collected and stored by the MDAS was available. If a person were hospitalized at an outside hospital, that hospitalization would not be represented in these analyses. Despite including travel time to the cancer center to address this limitation, and the finding that travel time was not a significant predictor of hospitalization, it is possible that this potentially missing hospitalization data could result in different findings.

Although conducted at a single academic, safety net, cancer center, which may limit generalization, this study benefitted from the large and diverse sample of cancer patient participants. The diversity across socioeconomic status, cancer diagnosis and stage, and race allow for greater generalizability of findings. Findings from this study also fill a gap in the current health literacy literature, as it is the first to examine the relationship between health literacy and hospitalization in a cancer patient population.

Conclusion

Using the CHLT-30, a psychometrically valid and reliable measure of cancer health literacy, this study identified a negative relationship between health literacy and number of inpatient admissions and total number of days hospitalized in a sample of cancer patients. This robust finding persisted after controlling for potentially confounding variables. This study considered only counts of hospitalization, and future work should focus on the causes of hospitalizations and extended hospital stays to gain a better understanding of the role of health literacy.

References

  1. Amalraj, S., Starkweather, C., Nguyen, C. & Naeim, A. (2009). Health literacy, communication, and treatment decision-making in older cancer patients. Oncology, 23(4), 369–375.
  2. Apter, A. J., Wan, F., Reisine, S., Bender, B., Rand, C., Bogen, D. K. & Gonzalez, R. (2013). The association of health literacy with adherence and outcomes in moderate-severe asthma. Journal of Allergy and Clinical Immunology, 132(2), 321–327. doi:10.1016/j.jaci.2013.02.014 [CrossRef]
  3. Baker, D. W., Parker, R. M. & Williams, M. V. (1998). Functional health literacy and the risk of hospital admission. Journal of General Internal Medicine, 13(12), 791–798. doi:10.1046/j.1525-1497.1998.00242.x [CrossRef]
  4. Bennett, I. M., Chen, J., Soroui, J. S. & White, S. (2009). The contribution of health literacy to disparities in self-rated health status and preventive health behaviors in older adults. The Annals of Family Medicine, 7(3), 204–211. doi:10.1370/afm.940 [CrossRef]
  5. Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J. & Crotty, K. (2011). Low health literacy and health outcomes: An updated systematic review. Annals of Internal Medicine, 155(2), 97–107. doi:10.7326/0003-4819-155-2-201107190-00005 [CrossRef]
  6. Brooks, G. A., Abrams, T. A., Meyerhardt, J. A., Enzinger, P. C., Sommer, K., Dalby, C. K. & Schrag, D. (2014). Identification of potentially avoidable hospitalizations in patients with GI cancer. Journal of Clinical Oncology, 32(6), 496–503. doi:10.1200/JCO.2013.52.4330 [CrossRef]
  7. Charlson, M. E., Pompei, P., Ales, K. L. & MacKenzie, C. R. (1987). A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal of Chronic Diseases, 40(5), 373–383. doi:10.1016/0021-9681(87)90171-8 [CrossRef]
  8. Davis, T. C., Dolan, N. C., Ferreira, M. R., Tomori, C., Green, K. W., Sipler, A. M. & Bennett, C. L. (2001). The role of inadequate health literacy skills in colorectal cancer screening. Cancer Investigation, 19(2), 193–200. doi:10.1081/CNV-100000154 [CrossRef]
  9. Dumenci, L., Matsuyama, R., Riddle, D., Cartwright, L., Perera, R., Chung, H. & Siminoff, L. (2014). Measurement of cancer health literacy and identification of patients with limited cancer health literacy. Journal of Health Communication, 19(2), 205–224. doi:10.1080/10810730.2014.943377 [CrossRef]
  10. Evangelista, L. S., Rasmusson, K. D., Laramee, A. S., Barr, J., Ammon, S. E., Dunbar, S. & Yancy, C. W. (2010). Health literacy and the patient with heart failure—Implications for patient care and research: A consensus statement of the Heart Failure Society of America. Journal of Cardiac Failure, 16(1), 9–16. doi:10.1016/j.cardfail.2009.10.026 [CrossRef]
  11. Friedman, D. B., Corwin, S. J., Dominick, G. M. & Rose, I. D. (2009). African American men's understanding and perceptions about prostate cancer: Why multiple dimensions of health literacy are important in cancer communication. Journal of Community Health, 34(5), 449–460. doi:10.1007/s10900-009-9167-3 [CrossRef]
  12. Hurwitz, L. M., Cullen, J., Elsamanoudi, S., Kim, D. J., Hudak, J., Colston, M. & Rosner, I. L. (2016). A prospective cohort study of treatment decision-making for prostate cancer following participation in a multidisciplinary clinic. Urologic Oncology: Seminars and Original Investigations, 34(5), 233.e17–233.e25. doi:10.1016/j.urolonc.2015.11.014 [CrossRef]
  13. International statistical classification of diseases and related health problems (9th Rev.). (1979). Geneva, Switzerland: World Health Organization.
  14. Kalichman, S. C., Pope, H., White, D., Cherry, C., Amaral, C. M., Swetzes, C. & Kalichman, M. O. (2008). Association between health literacy and HIV treatment adherence: Further evidence from objectively measured medication adherence. Journal of the International Association of Physicians in AIDS Care, 7(6), 317–323. doi:10.1177/1545109708328130 [CrossRef]
  15. Kim, S. P., Knight, S. J., Tomori, C., Colella, K. M., Schoor, R. A., Shih, L. & Bennett, C. L. (2001). Health literacy and shared decision making for prostate cancer patients with low socioeconomic status. Cancer Investigation, 19(7), 684–691. doi:10.1081/CNV-100106143 [CrossRef]
  16. Koay, K., Schofield, P. & Jefford, M. (2012). Importance of health literacy in oncology. Asia-Pacific Journal of Clinical Oncology, 8(1), 14–23. doi:10.1111/j.1743-7563.2012.01522.x [CrossRef]
  17. Lawless, J. F. (1987). Negative binomial and mixed poisson regression. Canadian Journal of Statistics, 15(3), 209–225. doi:10.2307/3314912 [CrossRef]
  18. Li, C. I., Malone, K. E. & Daling, J. R. (2003). Differences in breast cancer stage, treatment, and survival by race and ethnicity. Archives of Internal Medicine, 163(1), 49–56. doi:10.1001/archinte.163.1.49 [CrossRef]
  19. McNaughton, C. D., Kripalani, S., Cawthon, C., Mion, L. C., Wallston, K. A. & Roumie, C. L. (2014). Association of health literacy with elevated blood pressure: A cohort study of hospitalized patients. Medical Care, 52(4), 346–353. doi:10.1097/MLR.0000000000000101 [CrossRef]
  20. Miller, D. P., Brownlee, C. D., McCoy, T. P. & Pignone, M. P. (2007). The effect of health literacy on knowledge and receipt of colorectal cancer screening: A survey study. BMC Family Practice, 8(1), 16. doi:10.1186/1471-2296-8-16 [CrossRef]
  21. Mitchell, S., Sadikova, E., Jack, B. & Paasche-Orlow, M. (2012). Health literacy and 30-day postdischarge hospital utilization. Journal of Health Communication, 17(Suppl. 3), 325–338. doi:10.1080/10810730.2012.715233 [CrossRef]
  22. Muthén, L. K. & Muthén, B. O. (1998–2011). Mplus user's guide (6th ed.). Los Angeles, CA: Muthén & Muthén.
  23. Ojinnaka, C. O., Bolin, J. N., McClellan, D. A., Helduser, J. W., Nash, P. & Ory, M. G. (2015). The role of health literacy and communication habits on previous colorectal cancer screening among low-income and uninsured patients. Preventive Medicine Reports, 2, 158–163. doi:10.1016/j.pmedr.2015.02.009 [CrossRef]
  24. Omachi, T. A., Sarkar, U., Yelin, E. H., Blanc, P. D. & Katz, P. P. (2013). Lower health literacy is associated with poorer health status and outcomes in chronic obstructive pulmonary disease. Journal of General Internal Medicine, 28(1), 74–81. doi:10.1007/s11606-012-2177-3 [CrossRef]
  25. Paasche-Orlow, M. K. & Wolf, M. S. (2007). The causal pathways linking health literacy to health outcomes. American Journal of Health Behavior, 31(Suppl. 1), S19–S26. doi:10.5993/AJHB.31.s1.4 [CrossRef]
  26. Pagan, J. A., Brown, C. J., Asch, D. A., Armstrong, K., Bastida, E. & Guerra, C. (2012). Health literacy and breast cancer screening among Mexican American women in South Texas. Journal of Cancer Education, 27(1), 132–137. doi:10.1007/s13187-011-0239-6 [CrossRef]
  27. Parsons, H. M., Habermann, E. B., Stain, S. C., Vickers, S. M. & Al-Refaie, W. B. (2012). What happens to racial and ethnic minorities after cancer surgery at American College of Surgeons national surgical quality improvement program hospitals?Journal of the American College of Surgeons, 214(4), 539–547. doi:10.1016/j.jamcollsurg.2011.12.024 [CrossRef]
  28. Peterson, N. B., Dwyer, K. A., Mulvaney, S. A., Dietrich, M. S. & Rothman, R. L. (2007). The influence of health literacy on colorectal cancer screening knowledge, beliefs and behavior. Journal of the National Medical Association, 99(10), 1105–1112.
  29. Peterson, P. N., Shetterly, S. M., Clarke, C. L., Bekelman, D. B., Chan, P. S., Allen, L. A. & Masoudi, F. A. (2011). Health literacy and outcomes among patients with heart failure. The Journal of the American Medical Association, 305(16), 1695–1701. doi:10.1001/jama.2011.512 [CrossRef]
  30. Quan, H., Sundararajan, V., Halfon, P., Fong, A., Burnand, B., Luthi, J. & Ghali, W. A. (2005). Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical Care, 43(11), 1130–1139. doi:10.1097/01.mlr.0000182534.19832.83 [CrossRef]
  31. Ratzan, S. C. & Parker, R. M. (2000). Introduction. In: Selden, C.R., Zorn, M., Ratzan, S.C. & Parker, R.M. (Eds.), National Library of Medicine current bibliographies in medicine: Health literacy. Bethesda, MD: National Institutes of Health.
  32. Ravi, D. P., Sood, E. A., Schmid, O. M., Abdollah, S. F., Sammon, L. J., Sun, L. M. & Trinh, L. Q. (2015). Racial/ethnic disparities in perioperative outcomes of major procedures: Results from the national surgical quality improvement program. Annals of Surgery, 262(6), 955–964. doi:10.1097/SLA.0000000000001078 [CrossRef]
  33. Ross, L. E., Howard, D. L., Bowie, J. V., Thorpe, R. J., Kinlock, B. L., Burt, C. & LaVeist, T. A. (2016). Factors associated with men's assessment of prostate cancer treatment choice. Journal of Cancer Education, 31(2), 301–307. doi:10.1007/s13187-015-0837-9 [CrossRef]
  34. Scott, T. L., Gazmararian, J. A., Williams, M. V. & Baker, D. W. (2002). Health literacy and preventive health care use among medicare enrollees in a managed care organization. Medical Care, 40(5), 395–404. doi:10.1097/00005650-200205000-00005 [CrossRef]
  35. Seo, J., Goodman, M. S., Politi, M., Blanchard, M. & Kaphingst, K. A. (2016). Effect of health literacy on decision-making preferences among medically underserved patients. Medical Decision Making, 36(4), 550–556. doi:10.1177/0272989X16632197 [CrossRef]
  36. Shavers, V. L. & Brown, M. L. (2002). Racial and ethnic disparities in the receipt of cancer treatment. Journal of the National Cancer Institute, 94(5), 334–357. doi:10.1093/jnci/94.5.334 [CrossRef]
  37. Smith, S. K., Dixon, A., Trevena, L., Nutbeam, D. & McCaffery, K. J. (2009). Exploring patient involvement in healthcare decision making across different education and functional health literacy groups. Social Science & Medicine, 69(12), 1805–1812. doi:10.1016/j.socscimed.2009.09.056 [CrossRef]
  38. Sparks, L. & Nussbaum, J. F. (2008). Health literacy and cancer communication with older adults. Patient Education and Counseling, 71(3), 345–350. doi:10.1016/j.pec.2008.02.007 [CrossRef]
  39. White, S., Chen, J. & Atchison, R. (2008). Relationship of preventive health practices and health literacy: A national study. American Journal of Health Behavior, 32(3), 227–242. doi:10.5993/AJHB.32.3.1 [CrossRef]
  40. Williams, M. V., Baker, D. W., Parker, R. M. & Nurss, J. R. (1998). Relationship of functional health literacy to patients' knowledge of their chronic disease: A study of patients with hypertension and diabetes. Archives of Internal Medicine, 158(2), 166–172. doi:10.1001/archinte.158.2.166 [CrossRef]
  41. Wilson, E. A., Wolf, M. S., Curtis, L. M., Clayman, M. L., Cameron, K. A., Eigen, K. V. & Makoul, G. (2010). Literacy, cognitive ability, and the retention of health-related information about colorectal cancer screening. Journal of Health Communication, 15, 116–125. doi:10.1080/10810730.2010.499984 [CrossRef]
  42. Wu, J., Holmes, G., DeWalt, D., Macabasco-O'Connell, A., Bibbins-Domingo, K., Ruo, B. & Pignone, M. (2013). Low literacy is associated with increased risk of hospitalization and death among individuals with heart failure. Journal of General Internal Medicine, 28(9), 1174–1180. doi:10.1007/s11606-013-2394-4 [CrossRef]

Demographic Variables

Variable n (%)
Education
  High school diploma/GED or less 267 (35.6)
  More than a high school diploma 484 (64.4)
Insurance
  Publica 395 (52.6)
  Private 331 (44)
  Uninsured 25 (3.4)
Race
  Non-Hispanic White 458 (61)
  Non-Hispanic minority 293 (39)
Sex
  Female 377 (49.9)
  Male 375 (50.1)
Cancer diagnosis
  Hematologic 198 (26.3)
  Lung 90 (12)
  Head/neck 90 (12)
  Genitourinary 81 (10.8)
  Gynecologic 76 (10.1)
  Breast 76 (10.1)
  Colon/rectal/anal 56 (7.4)
  Gastrointestinal 34 (4.5)
  Skin 19 (2.5)
  Endocrine 10 (1.3)
  Other or unknown 22 (2.9)
Comorbidity diagnosis
  Metastatic carcinoma 257 (34.2)
  Chronic pulmonary disease 178 (23.7)
  Mild liver disease 161 (21.4)
  Diabetes without complications 161 (21.4)
  Congestive heart failure 72 (9.6)
  Renal disease 65 (8.6)
  Peptic ulcer disease 58 (7.7)
  Cerebrovascular disease 42 (5.6)
  Peripheral vascular disease 41 (5.5)
  Myocardial infarction 34 (4.5)
  Diabetes with complications 29 (3.9)
  AIDS/HIV 24 (3.2)
  Rheumatic disease 22 (2.9)
  Paraplegia or hemiplegia 17 (2.3)
  Moderate or severe liver disease 10 (1.3)
  Dementia 1 (0.1)

M1 CHLT-30 Multivariate Model Outcomes

Outcome Variable Range Mean (SD) Beta p Value
Number of inpatient hospital admissions 0–18 1.95 (2.55) −0.074 < .001
Total days hospitalized 0–145 10.81 (18.36) −0.024 .015
Number of 30-day readmissions 0–13 0.61 (1.56) −0.020 .213

M2 CHLT-30 Multivariate Model Outcomes

Predictors Number of Admissions Days Hospitalized Number of Readmissions CHLT-30
β p Value β p Value β p Value β p Value
CHLT-30 −0.029 .054 −0.021 .109 0.003 .871 - -
Stage IV −0.009 .962 −0.359 .023 −0.053 .826 0.055 .876
Education −0.284 .085 −0.197 .194 −0.162 .499 5.218 <.001
Race 0.202 .187 0.308 .041 0.143 .507 5.447 <.001
Chemotherapy 0.631 < .001 0.836 < .001 0.910 < .001 −0.345 .266
Hematology flag 0.743 < .001 0.665 < .001 1.227 < .001 −0.013 .969
Deceased 0.672 < .001 0.757 <. 001 0.542 .037 0.139 .755
Comorbidity count 0.337 < .001 0.325 <. 001 0.307 < .001 −0.281 .007
Log of time period 0.078 .553 0.004 .979 −0.091 .635 0.010 .976
Travel time 0.002 .124 0.003 .061 0.001 .826 −0.002 .750

M3 CHLT-30 Multivariate Model Outcomes

Predictors Number of Admissions Days Hospitalized Number of Readmissions CHLT-30
β p Value β p Value β p Value β p Value
CHLT-30 −0.041 .009 −0.028 .023 −0.002 .903 - -
Stage IV −0.011 .953 −0.365 .017 −0.056 .812 - -
Education - - - - - - 5.250 < .001
Race 0.254 .132 0.2331 .024 0.140 .532 5.426 < .001
Chemotherapy 0.630 < .001 0.823 < .001 0.913 < .001 - -
Hematology flag 0.696 < .001 0.646 < .001 1.242 < .001 - -
Deceased 0.571 .001 0.740 < .001 0.601 .003 - -
Comorbidity count 0.354 < .001 0.325 < .001 0.310 < .001 −0.283 .006
Authors

Laura A. Cartwright, PhD, is a Research Scientist, Department of Health Behavior and Policy, Virginia Commonwealth University. Levent Dumenci, PhD, is a Professor, Department of Epidemiology and Biostatistics, Temple University College of Public Health. J. Brian Cassel, PhD, is an Assistant Professor, Department of Internal Medicine, Division of Hematology/Oncology & Palliative Care, School of Medicine, Virginia Commonwealth University. Maria D. Thomson, PhD, is an Assistant Professor, Department of Health Behavior and Policy, Virginia Commonwealth University. Robin K. Matsuyama, PhD, is an Associate Professor, Department of Health Behavior and Policy, Virginia Commonwealth University.

Address correspondence to Laura A. Cartwright, PhD, Department of Health Behavior and Policy, Virginia Commonwealth University, 830 E. Main Street, Richmond, VA 23219; email: Laura.Cartwright@vcuhealth.org.

Disclosure: Laura A. Cartwright and Robin K. Matsuyama report grants from the National Institutes of Health/National Cancer Institute during the conduct of the study. The remaining authors have no relevant financial relationships to disclose.

This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (https://creativecommons.org/licenses/by-nc/4.0). This license allows users to copy and distribute, to remix, transform, and build upon the article non-commercially, provided the author is attributed and the new work is non-commercial.
Received: April 25, 2017
Accepted: August 09, 2017

10.3928/24748307-20170808-01

Sign up to receive

Journal E-contents