Cancer Survivors with Sub-Optimal Patient-Centered Communication Before and During the Early COVID-19 Pandemic.

Objectives Patient-Centered Communication (PCC) is an essential element of patient-centered cancer care. Thus, this study aimed to examine the prevalence of and factors associated with optimal PCC among cancer survivors during COVID-19, which has been less studied. Methods We used national survey (Health Information National Trends Survey) among cancer survivors (n=2,579) to calculate the prevalence (%) of optimal PCC in all 6 PCC domains and overall (mean) by time (before COVID-19, 2017-19 vs. COVID-19, 2020). Multivariable logistic regressions were performed to explore the associations of sociodemographic (age, birth gender, race/ethnicity, income, education, usual source of care), and health status (general health, depression/anxiety symptoms, time since diagnosis, cancer type) factors with optimal PCC. Results The prevalence of optimal PCC decreased during COVID-19 overall, with the greatest decrease in managing uncertainty (7.3%). Those with no usual source of care (odd ratios, ORs =1.53-2.29), poor general health (ORs=1.40-1.66), depression/anxiety symptoms (ORs=1.73-2.17) were less likely to have optimal PCC in most domains and overall PCC. Conclusions We observed that the decreased prevalence of optimal PCC, and identified those with suboptimal PCC during COVID-19. Practice implications More efforts to raise awareness and improve PCC are suggested, including education and guidelines, given the decreased prevalence during this public health emergency.


Patient Centered Communication (PCC) is defined as interactions and communications between patients and providers
to meet patients' needs and respond to their preferences [1]. The National Cancer Institute (NCI) outlined six core domains of PCC that could influence patients' essential health outcomes: exchanging information, responding to emotions, making decisions, enabling self-management, fostering healing relationships, and managing uncertainty [2].
PCC allows patients to have time with providers to ask questions and receive the relevant information to care for themselves, acquire support from the providers for health decision-making, and help to express emotions and deal with uncertainty and anxiety [3][4][5][6]. The Institute of Medicine (IOM) identified PCC as an essential element of patientcentered care in 2013 [7]. People who experienced PCC reported benefits from mental distress management [2,8].
They also showed higher cancer care quality, treatment adherence, emotional well-being, and health-related quality of life [9][10][11].
During the early COVID-19 pandemic in 2020, cancer survivors faced disrupted cancer care (e.g., delayed cancer care, changed treatment plans) and fear of disease progression [12][13][14][15]. After the unprecedented Stay At Home Order in March 2020 in United States (U.S.), in-person clinic visits were extremely limited. In addition, cancer survivors experienced additional fear of COVID-19 infection because those with chronic medical conditions, including cancer, showed worse COVID-19 infection outcomes [13,16]. The restricted in-person patient-provider interactions due to Stay At Home Order might have hindered optimal PCC during this time. In addition to this actual limitation of providers' communicational capacity (e.g., closed health care facility, lack of health providers), patients' perceived distance from providers due to the interaction-discouraged atmosphere during this unique time might have been also at play in preventing optimal PCC performance [13,15,17]. A study reported that physicians' responsiveness to patients during conversations to help address uncertain and difficult emotions was associated with better health and coping and less psychological distress during COVID-19 [17], highlighting the importance of PCC.
Prior studies have found PCC disparities by sociodemographic and health status factors among cancer survivors in the U.S. Cancer survivors who were racial/ethnic minorities, were more educated, had low income, had no usual source of care, or had poor physical or mental health reported lower perceived PCC [18][19][20][21], while age showed inconsistent associations. For example, older cancer survivors had higher perceived PCC in HINTS 4 (2011HINTS 4 ( -2013 [20], yet age J o u r n a l P r e -p r o o f was not related among newly diagnosed colon or rectal cancer patients [21]. Previously, Blanch-Hartigan et al.
assessed the trends in cancer survivors' PCC experience using HINTS 2007-2013 [19]. However, a systematic evaluation of all six PCC domains among cancer survivors during the early pandemic has not been conducted, limiting our ability to examine the impact of COVID-19 pandemic overall PCC performance and across the subgroups.
Therefore, this study used the nationally representative HINTS data (2017 to 2020) to assess the prevalence of optimal PCC, defined as always having perceived PCC [20], among cancer survivors during COVID-19 compared to those before COVID-19. This study also investigated sociodemographic and health status characteristics associated with optimal PCC during COVID-19 to identify subgroups of cancer survivors who would need support to have optimal PCC. We hypothesized that the prevalence of optimal PCC would decrease during the pandemic and the subgroups of cancer survivors with sub-optimal PCC would differ during COVID-19 than before COVID-19. Findings from this study can inform targeted interventions to support those in need. Furthermore, the knowledge could also contribute to improving PCC during telehealth visits that became rapidly and widely implemented during COVID-19 [22].

Data source
We used nationally representative survey data from Health Information National Trends Survey (HINTS) for this study [23]. HINTS is a self-administered, publicly available, cross-sectional survey data distributed and collected by National Cancer Institute (NCI) [24]. This study used the HINTS 5 data, Cycles 1-3 (2017-2019) for before COVID- 19 and Cycle 4 (2020) for during COVID-19. Of note, the COVID-19 sample was collected from February to June 2020. The survey questionnaires were administered to non-institutionalized civilians 18 years and older in the United States. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines to report for an observational study [25] (Supplemental Table 1). The total number of survey responses in HINTS 5 Cycles 1-4 was 16,092 and the average response rate was 33% [26]. Among the total responses, those with a history of cancer diagnosis were designated as cancer survivors' responses (n=2,579) in this study as we followed the NCI definition of cancer survivor, a person with cancer from the time of diagnosis until the end of life [27]. The HINTS reconciled the data from the different survey modes (mailed, push-to-web with a paper return, push-to-web with web return). We examined our variables of interest before combining 4 survey cycles to make sure the variable names and J o u r n a l P r e -p r o o f codes were consistent across the cycles. We used the HINTS Data Merging Code Tool that the HINTS provides to merge the data of HINTS 5 Cycle 1 to Cycle 4 [28]. We obtained 200 replicate weights and used those to calculate standard errors. The full-sample weights were applied for the data to be nationally representative, intending to account for household-level base weight, non-response, and person-level initial weight [29].

Outcomes
PCC was defined by the NCI framework [2] and measured using the following questions: "In your communication with all doctors, nurses, or other health professionals in the past 12 months, how often did they 1) give you a chance to ask health questions? (Exchanging information), 2) had the attention you needed to your feelings and emotions?
(Responding to emotions), 3) involve you in decisions about your health care as much as you wanted? (Making decisions), 4) make sure you understood the things you needed to do to take care of your health? (Enabling selfmanagement), 5) explain things in a way you could understand? (Enabling self-management), 6) spend enough time with you? (Fostering healing relationships), 7) help you deal with uncertain feelings about your health or health care? (Managing uncertainty)." Responses for each question were measured on a Likert scale (1=always, 2=usually, 3=sometimes, 4=never). As done previously with HINTS data [19,20], overall PCC was analyzed as a dichotomous outcome when all 6 domains were "always" for optimal PCC. Given this stringent cut-off, responses were combined and recoded using the Likert scale numbers to generate a new continuous PCC outcome variable, ranging from score 0 (the least optimal, when all 6 domains were scored "never") to score 100 (the most optimal, when all 6 domains were scored "always") to allow for comparisons to prior studies [19,20]. Furthermore, we dichotomized response options of each of the 6 domains as optimal (always) vs. sub-optimal (usually, sometimes, never) for our analysis and a sensitivity analysis was done to assess if the different cut-points [optimal (always/usually) vs. sub-optimal (sometimes/never)] would affect the associations, consistent with prior work [19].

Sociodemographic characteristics
We chose sociodemographic factors as independent variables of this study based on the social determinants of health conceptual framework from the Healthy People 2030 [30]: age, birth gender, race/ethnicity, household income, educational attainment, marital status (married or living with a romantic partner as a married vs. not married including divorced, widowed, separated, single/never been married), employment status (employed vs. unemployed including J o u r n a l P r e -p r o o f homemaker, student, retired, disabled), health insurance type, usual source of care, and rurality of residence (metropolitan, micropolitan, small town, rural). HINTS used Urban Rural Commuting Area (RUCA) to designate the rurality of residence of the survey respondents, which categorized census tracts using population density, urbanization, and commuting patterns developed by the United States Department of Agriculture [31].

Health status characteristics
Health status factors included general health status (excellent, very good, good, fair, poor), chronic medical conditions (diabetes, high blood pressure, heart disease, lung disease, depression), time since cancer diagnosis (less than a year, 2-5 years, 6-10 years, more than 11 years), cancer type (breast, cervical, prostate, colon, lung, skin cancer, melanoma, other cancer, or multiple cancer), and measures of psychological distress (little interest, hopelessness, nervousness, worrying). The psychological distress measurements were converted to depression or anxiety symptoms (past 2 weeks) using Patient-Health Questionnaire-4 (PHQ-4), and following its clinical cut-off (score ≥3, then symptom presents) [32].

Statistical analysis
Weighted descriptive analyses [percentage with standard error (SE)] was conducted to describe cancer survivors' sociodemographic and health status characteristics. To assess the prevalence of optimal PCC for each of the 6 domains and overall [dichotomized response (optimal = the response was 'always')] by time period (before and during COVID- 19), we calculated the weighted percentage (%) with SE. Additionally, to examine the overall continuous PCC by sociodemographic and health status factors over the entire study period and in before and during COVID-19 time periods, we calculated the overall mean PCC and SE.
To investigate the factors associated with optimal PCC (optimal=the responses of each domain was 'always'), multivariable-adjusted weighted logistic regression models were developed to estimate the odds ratio (OR) and 95% confidence intervals (95% CI) of optimal PCC using dichotomized response for each domain. The same model was applied for a dichotomous overall PCC (optimal=the responses of all 6 domains were 'always'). To explore the factors associated with a continuous overall PCC score, a multivariable-adjusted weighted linear regression model was developed to obtain coefficients () with SE. Sociodemographic and health status variables for the logistic and linear J o u r n a l P r e -p r o o f regression models, included age, gender, race/ethnicity, education, income, usual source of care, general health status, depression or anxiety symptoms, time since diagnosis, and cancer type. These variables were retained in the final model because they were considered as confounders (e.g., the covariate effect estimate changed by more than 10%), significantly associated with the outcome in univariable models (P<0.05) or were associated with PCC in prior studies [19,20,33]. To investigate whether the PCC differed during COVID-19 compared with before COVID-19, we assessed the associations between selected sociodemographic and health status factors (age, income, gender, usual source of care, race/ethnicity, and depression/anxiety symptoms) and time period (before vs. during COVID-19) in each model. Sensitivity analysis was conducted with [optimal (always/usually) vs. sub-optimal (sometimes/never)] to investigate the associated factors further by domain, as done previously [19].
We assessed the interactions of selected sociodemographic and health status factors (age, income, gender, usual source of care, race/ethnicity, and depression/anxiety symptoms) with time period (before vs. during COVID-19) with overall PCC score. For these interaction assessments, we included interaction terms in multivariable weighted linear regression models. We performed hot deck imputation, which the HINTS used to account for the non-response [29], to account for the missing data in the covariates, which ranged from 1.0% to 13.3% (see footnotes of Table 1). For all descriptive and regression analyses, the imputed data were used in SAS 9.4 (SAS studio 3.8, Cary, NC, USA). The statistical significance was determined at a P < 0.05.

Cancer survivor characteristics
In HINTS 5 2017-2020, there were 2,579 cancer survivors, 75% before (n=1,953) and 25% during COVID-19 (n=626) time periods (Table 1). About half (51%) were older adults (≥65 years), non-Hispanic Whites were the majority (80%), 66% had some college education or more, more than half (53%) reported $50,000 or more income, 57% had public/government-supported insurance, 84% had a regular provider, and 75% rated their health status as excellent/good. High blood pressure (54%) was the most common co-morbid chronic condition, followed by diabetes (24%) and depression (23%). Nearly one in three cancer survivors reported depression or anxiety symptoms in the past 2 weeks (33%). Almost half have been cancer survivors for more than 11 years (47%). There were no significant J o u r n a l P r e -p r o o f differences in population characteristics of cancer survivors between before and during COVID-19 (Supplemental Table 2). However, none of these differences were statistically significant.   Table 2 shows the mean PCC by sociodemographic and health status factors over the entire study period and before and during COVID-19 (higher score refers to better PCC). The PCC mean score significantly differed in some sociodemographic subgroups by time. From before to during COVID-19, the PCC mean score increased in non-Hispanic Black/African Americans and decreased in those in the middle-income bracket ($50,000 to < $75,000).

Impact of COVID-19
We did not observe interactions between COVID-19 time period and sociodemographic or health status factors with overall PCC score. Thus, associations of sociodemographic and health status factors with optimal PCC before and during COVID-19 were combined in Table 3 and Supplemental Table 3.

Factors associated with optimal PCC in each domain and overall
Compared with the time before COVID-19, cancer survivors during COVID-19 were less likely to have optimal PCC overall (OR=0.73, 95% CI 0.54-0.98) and in the domain of managing uncertainty (OR=0.74, 0.55-0.99) ( Table 3).
However, other PCC functions did not differ between the two time periods. Cancer survivors who had a usual source of care were 1.5-2 times as likely to have optimal PCC than those without it overall (OR=1. 53 Individuals diagnosed with cancer more recently (2-5 years ago) had a higher odds of having optimal PCC (ORs=1.51-1.53) in exchanging information and enabling self-management than those diagnosed 11 years ago.
In the linear regression models considering overall PCC score, most associations were similar to optimal PCC, with the exception that the COVID-19 time period was not significantly related to the overall PCC score (Supplemental Table 3). Sensitivity analysis revealed that the associations remained the same for most sociodemographic and health status factors, except for gender. Gender was not associated with PCC outcomes when 'always/usually' were treated as optimal PCC. J o u r n a l P r e -p r o o f (7.3% lower) domains, yet this unadjusted prevalence did not significantly differ. However, in the fully adjusted model, cancer survivors during COVID-19 were less likely to have optimal PCC overall and in managing uncertainty than cancer survivors before pandemic. Substantially limited interactions with providers (actual restriction) and pervasive social distancing policies (perceived distance from providers) might have prevented them from having quality communications for managing uncertainty despite their elevated fear and uncertainties. Before COVID-19, several efforts to enhance the quality of PCC have been put into the practice [34], including educational PCC training for healthcare providers (e.g., family physician residents, nursing students) [35,36] and attempts to improve PCC assessment tools (e.g., standardization and validation of PCC check list, engagement of patient advocates to improve PCC design and content) [37,38]. However, systematic PCC practice guidelines and evaluations for clinicians, and consistent and broadly available education for patients are still lacking. Despite those previous efforts, the prevalence of optimal PCC and some domains have decreased over time, even lower than estimated from a study during 2008-2013 [19], highlighting the need to focus more attention and resources on promoting PCC.
We observed that having depression or anxiety symptoms or poor general health status were consistently associated with sub-optimal PCC in most PCC domains among cancer survivors, which aligns with previous reports [18][19][20].
While PCC is ideal at all times, under the situations like COVID-19 pandemic, when individuals with compromised health conditions, including cancer patients, experienced additional fear due to COVID-19, the PCC's role is crucial as a channel to address those uncertainties and receive necessary care and support. It is possible that those with poor health status were less likely to be engaged in the communications with providers (e.g., disinterested or unable to), and the providers were also less likely to be patient-centered for those less attentive during the communication [20,21]. Our findings highlight the importance of preparing targeted approaches for those with poor physical or mental health to improve PCC, which has been found to be positively related to better health-related outcomes, including disease outcomes, quality of life, and mental health [9,11,39].
In our study, those without a usual source of care were less likely to have optimal PCC in most domains, as found previously [19]. This finding may relate to consistent medical encounters enabling quality patient-physician relationships and positively impacting optimal PCC [33,40]. Previously, cancer survivors with low-income were less likely to have optimal PCC, and had a higher rate of discontinuation of treatment or disease care [41][42][43], which may relate to inconsistent or less frequent office visits due to financial barriers. However, in our study, those in the middleincome bracket ($50,000 to <$75,000) had a lower likelihood of having optimal PCC, compared with those with the lowest income in the emotion, relationship, and uncertainty domains. We also observed that the overall PCC score significantly worsened in this middle-income bracket during COVID-19 compared to before the pandemic. Further investigations, including qualitative approaches, are warranted to understand the underlying dynamics in this observation. It is notable that 16% of cancer survivors did not have a usual source of care and 19% were middle income in our study.
Cancer survivors 65-74 years-old had a higher likelihood of optimal PCC than those 75 years of age and older in the enabling self-management and making decisions domains. The 75+ group may prefer to have strong control in care, asking direct questions, refusing some treatment options, or valuing 'being understood' during the communication [33,44]. Demanding more quality in care among the oldest could potentially contribute to less satisfactory PCC in these patient involvement related domains. More than half of cancer survivors (53%) were age 70 or older in the U.S. in 2022, and it is projected to be growing [45]. Thus, our findings indicate that more resources will need to be put into the oldest group to support them to achieve optimal PCC. Perhaps, national efforts for healthy aging could potentially incorporate opportunities to inform and educate older adults to improve PCC [46,47].
Male cancer survivors were consistently less likely to have optimal PCC than females in most domains. It aligns with the previous literature, which reported that male cancer survivors experienced sub-optimal PCC in managing uncertainty [19]. This may reflect gender differences in communicational styles, as women are more likely to share their issues or concerns with providers than men [34,48]. Typically, care providers can be more informative and supportive when they better understand patients' issues [49]. Given the gender gap in optimal PCC widened in recent years, further investigations to understand the underlying reasons for PCC differences are warranted.

Practice implications
To improve PCC among the vulnerable cancer survivors identified in this study, educational programs and guidelines/ policies for both healthcare providers and patients are suggested to raise awareness of PCC roles for both groups and guide them to practice PCC in clinical settings [50][51][52]. For example, early-stage trainings could be offered to health J o u r n a l P r e -p r o o f professionals on performing PCC and identifying vulnerable subgroups, particularly those with poor general health or mental health symptoms [35,36,38]. Moreover, patient advocate groups for the older or male cancer survivors could play roles in tailored patient education. Also, healthcare providers are usually more responsive to the patients who ask questions and share concerns, like other social interactions [53]. Additionally, there is evidence that racial/ethnic provider-patient concordance could facilitate positive interactions and relationships [54]. Last, exploring opportunities to enhance optimal PCC through online platforms (e.g., communications using Electronic Health Record to increase patients engagement) are timely with the widespread use of digital devices [55,56]. Online platforms could reach broad populations, including those without a usual source of care. Furthermore, given the rapid adoption and wide dissemination of telehealth during the pandemic, efforts may need to focus on engaging clinicians with PCC in telehealth services [22,57,58].

Limitations
Some limitations need to be acknowledged. First, the present study used self-reported survey data. Although the HINTS is a nationally representative, high-quality dataset, there is the possibility of reporting bias (e.g., some PCC responses could be reported subjectively, including the 'spending enough time with you' question because the same amount of time could be enough for some and not for others). Second, the possibility of selection bias needs to be acknowledged due to low overall response rate (33%, 2017-2020). Third, because the data are cross-sectional, we were not able to determine the prospective and longitudinal associations with optimal PCC. Fourth, the COVID-19 data were collected from February to June 2020, during the early COVID-19 pandemic. Hence, the findings should be interpreted in the early COVID-19 pandemic context, and the findings may differ in later or post COVID-19 periods.
Despite the limitations, this study has strengths, including the comprehensive investigations of the prevalence and associations by sociodemographic and health status factors with the optimal PCC by domains as well as the overall PCC with recent data, including in the context of the COVID-19 pandemic on a population level. This information contributes to our knowledge base of the PCC performance of vulnerable populations with chronic conditions, like cancer, during COVID-19.

J o u r n a l P r e -p r o o f
Our findings highlight that cancer survivors without a usual source of care, with depression or anxiety symptoms or with poor general health status, or those who were older, males, non-Hispanic Whites, or had middle-income require additional support to achieve optimal PCC during the extended COVID-19 pandemic. Raising awareness of PCC roles among both providers and cancer survivors and guiding them to practice it are suggested strategies to improve PCC.
The knowledge generated by this study informs related stakeholders, including healthcare professionals, public health professionals and policymakers, of the subgroups of cancer survivors to target with approaches to improve PCC performance and potentially prevent further disparities in health outcomes in these vulnerable populations.

Statement & Declaration
• Funding J o u r n a l P r e -p r o o f Theresa H. Keegan's time was supported, in part, by the UC Davis Comprehensive Cancer Center and National Cancer Institute of the National Institutes of Health under award number P30CA093373. Melanie S. Dove was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant number UL1 TR001860 and linked award KL2 TR001859. However, these funding sources did not have any roles in study design, in the collection, analysis and interpretation of data, in the writing of the report, and in the decision to submit the article for publication.

• Data availability
Current study used publicly available national survey data (Health Information National Trends Survey, HINTS 5 2017-2020). The raw data required to reproduce the above findings are available to download from https://hints.cancer.gov/data/download-data.aspx The processed data required to reproduce the above findings will be made available upon request.

• Ethics approval
This study was a secondary data analysis. Human subject was not involved, and identifiable information was not included. Thus, this was deemed exempt for review by the Institutional Review Board at University of California, Davis • Acknowledgement N/A