Patient Perceptions of the Impact of the COVID Pandemic on the Quality of Their Gastrointestinal Cancer Care

Abstract

We surveyed patients who had a received care for a gastrointestinal cancer between 03/2020 and 05/2021 to understand their perceptions of the impact of the Covid pandemic on cancer care delivery and quality of care. Three-hundred fifty-eight respondents provided evaluable responses (response rate: 17.3%). Approximately half of respondents (46.4%) perceived that they had experienced a pandemic-related cancer care modification; most changes were initiated by a clinician or the cancer center (44.6%). Relative to White patients those from Racialized Groups (OR: 1.91, 95% CI: 1.03-3.54) were more likely to report a cancer treatment change. Additionally, relative to patients in follow-up, those who were newly diagnosed (OR: 2.39; 95% CI: 1.21-4.71) were more likely to report a change. Compared to White patients, patients from Racialized Groups were approximately twice as likely to report perceiving that virtual visits during Covid negatively impacted the quality of their care (OR: 2.21; 95% CI: 0.96-5.08). These findings potentially reflect pre-existing systemic disparities in quality of and access to care, as well as differences in how care is experienced by patients from Racialized Groups.

Introduction

Beyond Covid-19 (Covid) infection itself, the pandemic has impacted how some cancer care is delivered. ¹ Many articles at the outset of the pandemic indicated that delays and cancellations were often patient-initiated out of fear of contracting Covid, ¹ though some changes were implemented by institutions to facilitate delivery of care to patients with Covid, such as reducing the number of scheduled surgical resections to decrease the need for intensive care unit capacity and ventilators,^1-3 while others were introduced to mitigate transmission, such as reducing the number of in-person visits by prioritizing of curative intent treatments, restricting the numbers and/or types of essential care partners attending visits, and using oral rather than intravenous therapies.^4-7 The extent to which cancer care was modified in response to the pandemic and its relationship to clinical outcomes is not yet well understood; however, changes to care delivery have been linked to substantial delays in some jurisdictions, particularly for patients who were newly diagnosed, ⁸ were frail or had incurable disease. ⁹

The pandemic may also have exacerbated pre-existing disparities in care by disproportionately impacting sociodemographic subgroups who experience delayed access to screening or diagnostic services, and have higher stage at presentation under non-pandemic conditions. A recent survey of adult patients in the United States treated during the pandemic found that relative to White respondents, Black respondents (OR: 6.13; 95% CI: 3.50-10.74) and Hispanic or Latinx respondents (OR: 2.77; 95% CI: 1.49-5.14) were significantly more likely to experience involuntary treatment delays, and that Black respondents were 3 times as likely to experience a delay greater than 4 weeks (OR: 3.13; 95% CI: 1.11-8.81). ¹⁰ While virtual visits were widely adopted at the beginning of the pandemic in an effort to maintain levels of care and reduce the number of in-person visits, this may have also had implications for care quality, as lower engagement with, and negative perceptions of virtual care have been reported for older adults ¹¹ and patients in Racialized Groups.^12,13

Some practice changes brought about during the pandemic, such as the implementation of virtual care and reduction of use of some laboratory tests and imaging,^14,15 are likely to be sustained beyond the pandemic. As such, there is a need to examine the impact of the pandemic on quality of oncologic care. Herein, we undertook a cross-sectional survey of patients who received care for a gastrointestinal cancer during Covid at the largest comprehensive cancer center in Ontario, Canada to evaluate patients’ perceptions of the pandemic's impact on cancer care delivery and quality of care. Prior to the pandemic, approximately 0.8% of daily visits at our center were conducted virtually (either by video or telephone) which rose rapidly to 68.4% during the early stages of the pandemic, and still represents approximately 1 in 4 visits at our cancer center. ¹⁶ As such, we included questions specifically probing patient perceptions around the impact of virtual visits on quality of care, and the patient-centeredness quality domain. In addition, we undertook regression analyses to evaluate potential associations with disease characteristics and social determinants of health to examine patient-level differences in experiences with Covid-related modifications to cancer care and negative perceptions of quality of care.

Methods

Setting and Ethics

This work was undertaken at Princess Margaret Cancer Centre (PM), a large, urban comprehensive cancer center in Toronto, Canada where health services are provided through a single-payer, universal health care system; the gastrointestinal cancer clinic sees approximately 2900 patients annually. The study was approved by the University Health Network (#21-5276), and the University of Toronto (#41345) Research Ethics Boards.

Survey Content

As an existing validated measure could not be identified to appropriately capture modifications to cancer care, a survey measure was constructed which captured patient perceptions of the impact of the pandemic on treatment decision making and the underlying reasons for those decisions, informed by a scoping literature review (Supplemental File 1). ¹ We quantified patients’ perception of the quality of care that they received based on the National Academy of Medicine's definition of patient-centered quality,^17,18 and the Picker Institute's 8 principles of patient-centered care (respect for patients’ preferences, coordination and integration of care, information and education, physical comfort, emotional support, involvement of family and friends, continuity and transition, and access to care) on a 5-point Likert scale from 1-never to 5-always. ¹⁹ Taking into account previous studies suggesting that patient overall quality ratings and willingness to recommend care are most strongly tied to technical performance on established quality metrics, ²⁰ we adapted the wording from the provincially-mandated Cancer Care Ontario/Ontario Health Ambulatory Patient Experience Survey ²¹ to query patients on their overall impressions of their care. Categorical sociodemographic and clinical characteristics collected from respondents were selected based on previous reports evaluating the impact of patient-level characteristics on patient experience, ²² and on the PROGRESS (Place of residence, Race/ethnicity/culture/language, Occupation, Gender/sex, Religion, Education, Socioeconomic status and Social capital) equity framework. ²³ For respondents ≥65 years of age, frailty was captured using the validated Vulnerable Elders Survey-13, ²⁴ and dichotomize to not frail (0-2) or frail (3+).

Survey Dissemination and Data Collection

All patients ≥18 years of age who had a visit in the gastrointestinal clinic at PM between March 1, 2020 and May 31, 2021, and had a valid email address on file (approximately 60%-70% across PM) were invited to participate in the survey. Patients who required help from a caregiver to complete the survey were included. An introductory email was sent to eligible participants with a link to the consent and survey; by accepting the terms of the consent information form, participants provided consent to participate in the survey. As responses were anonymous, a reminder email was sent to all potential participants approximately 2 weeks later to maximize response rates for analysis. ²⁵ Survey responses were collected using the secure, SSL-enabled QuestionPro platform (QuestionPro; Dallas, USA).

Statistical Analysis

All analyses were undertaken using R (version 4.1.1, R Foundation for Statistical Computing; Vienna, Austria). Cronbach's alpha was calculated for questions in the panel pertaining to patient centeredness of care, which demonstrated moderately high covariance among items relative to the variance (Cronbach’s alpha: 0.89). Descriptive statistics were utilized to summarize patient and disease characteristics, and responses to individual survey questions. Patients were considered to be “in follow-up” if they indicated that they had completed treatment and were being followed for disease progression or recurrence. Patient status as existing or new patients at the start of the pandemic are relative to March 2020 as that is when the World Health Organization declared Covid to be a pandemic, and many of the pandemic-related changes to cancer care were enacted. Unadjusted and adjusted associations with patient-level demographic and clinical factors and reporting (1) a Covid-related modification to care, or (2) a negative perception of quality, were evaluated using separate univariate and multivariable binary logistic regression models. To build a parsimonious model, covariates for multivariable models were selected using forward stepwise selection by Akaike Information Criterion. ²⁶ Results were reported as odds ratios with 95% confidence intervals; statistical significance was defined as P < .05. To examine potential impact of representativeness of the survey sample, iterative proportional weighting was used to generate weight estimates relative to the Toronto, Canada population parameters for ethnicity, language and education, ²⁷ trimmed to 0.3-3.0. ²⁸ As weighting did not change the direction or conclusions of the analyses, all analyses are reported based on the unweighted responses.

Results

Cohort Characteristics

The response rate was 17.3% (435/2515); 358 responses were evaluable (Supplemental Figure 1). The majority of respondents were patients themselves rather than a caregiver on behalf of the patient (89.7%, Table 1. The greatest proportion of respondents were ≥65 years of age (46.6%), self-reported to be White (63.4%), spoke English as their first language (67.3%), were referred to PM prior to the start of the Covid pandemic (56.4%), and were in long-term follow-up for their cancer (42.2%; 151/358).

Table 1.

Self-reported Demographic and Clinical Characteristics of Respondents (n = 358).

Variables		N (%)
Respondent type	Patient	321 (89.7)
	Caregiver on behalf of patient	36 (10.0)
	Missing	1 (0.3)
Age	20-34	6 (1.7)
	35-44	13 (3.6)
	45-54	38 (10.6)
	55-64	75 (20.9)
	65-74	111 (31.0)
	75-84	47 (13.1)
	>85	9 (2.5)
	Missing	59 (16.5)
Gender	Female	127 (35.5)
	Male	174 (48.6)
	Other	2 (0.6)
	Missing	55 (15.4)
Racea	White	227 (63.4)
	Racialized Groups	76 (21.2)
	Missing	55 (15.4)
Language	English	241 (67.3)
	Other	61 (17.0)
	Missing	56 (15.6)
Education	Grade school	9 (2.5)
	Some high school	6 (1.7)
	High school	27 (7.5)
	Some college/ university	61 (17.0)
	University	125 (34.9)
	Graduate school	70 (19.6)
	Prefer not to answer	3 (0.8)
	Missing	57 (15.9)
Income	<$50 000	61 (17.0)
	$50 000-$75 000	48 (13.4)
	$75 000-$100 000	37 (10.3)
	>$100 000	96 (26.8)
	Prefer not to answer	59 (16.5)
	Missing	57 (15.9)
Marital status	Single	25 (7.0)
	Married/common-law	216 (60.3)
	Divorced/ widowed	52 (14.5)
	Prefer not to answer	9 (2.5)
	Missing	56 (15.6)
Frailty score b	0-2	105 (62.9)
	3+	62 (37.1)
Phase of cancer trajectory	Newly diagnosed	71 (19.8)
	Recurrent disease	102 (28.5)
	In follow-up c	151 (42.2)
	Missing	34 (9.5)
Referral to cancer center	Prior to the start of Covid d	202 (56.4)
	During Covid	150 (41.9)
	Missing	6 (1.7)

^a Patients who self-reported belonging to a race other than White.

^b Restricted to patients ≥65 years of age.

^c Patients who indicated that they had completed treatment and were still being followed for disease progression/ recurrence.

^d Covid start was considered March 2020.

Covid-related Modifications to Care

Approximately half of the respondents (46.4%) indicated that they felt the Covid pandemic had resulted in some change to their cancer care. The most frequently reported modifications to care were change, delay or deferral of in-person appointments with their oncologists (68.7%), advanced imaging scans (28.9%), and surgical procedures (24.1%; Table 2. Most modifications to care were reportedly initiated by a clinician or the cancer center (44.6%), while few patients reported that they had requested a change (5.4%). Of those patients who did request a change in their care (n = 9), they most frequently did so due to fear of being infected with Covid during their appointment, or due to issues with travelling to the cancer center (data not shown). Relative to White respondents, patients from Racialized Groups were almost twice as likely to report having experienced a Covid related change to their imaging scans, genetic testing or receipt of cancer treatment (OR: 1.91, 95% CI: 1.03-3.54; Table 3. Relative to patients in follow-up care, patients who were newly diagnosed were more than twice as likely to report having experienced a Covid-related change to their care (OR: 2.39; 95% CI: 1.21-4.71). Frailty was not found to be associated with experiencing a Covid-related change in care (P = .63).

Table 2.

Summary of Responses of Patients who Reported Having Experienced a Modification to Care During Covid (46.4%; 166/358).

Modification to care		N (%)
Experienced a change, delay or deferral of cancer care a	In-person appointments with oncologist	114 (68.7)
	Imaging services (MRI, CT scan, PET)	48 (28.9)
	Genetic testing	6 (3.6)
	Supportive services (physical therapy, psychosocial)	16 (9.6)
	Surgical procedures	40 (24.1)
	Radiotherapy	11 (6.6)
	Systemic therapy	19 (11.4)
Impact to treatment initiation	Delayed by less than 2 weeks	8 (4.8)
	Delayed by more than 2 weeks, but less than 3 months	20 (12.0)
	Delayed by more than 3 months	4 (2.4)
	Delayed by more than 3 months, but less than 6 months	4 (2.4)
	Delayed by more than 6 months	3 (1.8)
	Delayed, not yet rescheduled	3 (1.8)
	No change, appointments carried out as planned	104 (62.7)
	Missing	20 (12.0)
Impact to cancer treatment delivery a	Change of treatment modality (ie, surgery vs concurrent systemic therapy—radiotherapy)	5 (3.0)
	Change of treatment administration (ie,: oral vs intravenous systemic therapy)	4 (2.4)
	No change	111 (66.9)
	I am unsure	31 (18.7)
Impact to visits with oncologist	Some visits by telephone	71 (42.8)
	More than half visits telephone	18 (10.8)
	All visits by telephone	24 (14.5)
	Some visits by video	12 (7.2)
	More than half visits by video	3 (1.8)
	All visits by video	3 (1.8)
	No change, all visits in person	16 (9.6)
	Missing	19 (11.4)
Reason for change, delay or deferral of cancer care	Patient requested	9 (5.4)
	Awaiting Covid test results	4 (2.4)
	Provider/institution initiated	74 (44.6)
	Other	6 (3.6)
	Missing	73 (44.0)

^a Respondents could select more than one.

Table 3.

Unadjusted and Adjusted Associations Between Patient-level Demographic and Clinical Factors and Reporting a Covid-related Modification to Receipt of Imaging Tests, Genetic Testing, or Cancer Treatment Based on Univariate and Multivariable Logistic Regression.

		Univariate			Multivariable
Variables		Odds ratio	95% CI	P-value	Odds ratio	95% CI	P-value
Respondent type	Patient	Ref	–	.75
	Caregiver	1.13	0.52-2.46
Age	20-34	3.44	0.65-18.11	.74
	35-44	2.15	0.65-7.16
	45-54	1.07	0.45-2.55
	55-64	1.11	0.55-2.21
	65-74	Ref	–
	75-84	1.18	0.53-2.61
	>85	<0.01	9.30e-25-4.38e17
Gender	Male	Ref	–	.70
	Female	1.06	0.62-1.81
	Other	3.24	0.20-53.02
Race	White	Ref	–	.01	Ref	−	.04
	Racialized groups	2.16	1.22-3.82		1.91	1.03-3.54
Language	English	Ref	–	.16
	Other	1.56	0.84-2.90
Income	<$50 000	1.95	0.94-4.06	.44
	$50 000-$75 000	1.05	0.45-2.49
	$75 000-$100 000	1.48	0.61-3.58
	>$100 000	Ref	–
	Prefer not to answer	1.24	0.57-2.72
Education	Grade school	2.08	0.49-8.93	.33
	Some High School	2.08	0.36-12.05
	High school	1.46	0.55-3.84
	Some college	1.74	0.86-3.54
	University	Ref	–
	Graduate school	1.44	0.72-2.90
	Prefer not to answer	2.08	0.18-23.94
Marital status	Single	1.25	0.49-3.16	.88
	Married/common-law	Ref	–
	Divorced/widowed	0.97	0.47-1.98
	Prefer not to answer	1.61	0.39-6.66
Frailty score	0-2	Ref	–	.63
	3+	1.20	0.56-2.53
Phase of cancer trajectory	Newly diagnosed	2.48	1.33-4.62	.02	2.39	1.21-4.71	.04
	Recurrent disease	1.65	0.92-2.97		1.17	0.59-2.33
	In follow-up	Ref	–		Ref	–
Referral to cancer center	Prior to the start of Covid	Ref	–	.05
	During Covid	1.61	1.00-2.59

Patient Perceptions of Cancer Care Quality

Perceptions were positive on most domains of patient-centeredness (Supplemental Figure 2), particularly with respect to being provided information on tests (always: 83.6%) and treatments (always: 83.8%), and being adequately involved in decisions about their care (always: 71.6%). However, of the patients requiring interpretation services (n = 44), only 36.4% reported the services being routinely offered. Approximately one-third of respondents reported receiving inadequate emotional supports, with 36.0% and 31.8% of respondents reporting receiving inadequate access to information on emotional supports available to them and receipt of appropriate resources, respectively. Approximately one-fifth of respondents reported that they were not adequately informed of changes to their treatments or care (20.9%; 48/230). Similarly, 21.4% of respondents (51/238) felt that their friends and family had not been involved in decisions about their care as much as they preferred.

The majority of respondents rated the overall quality of their cancer care at PM as excellent or very good (89.2%), and indicated that they were very likely (93.1%) to recommend PM to family or friends (Figure 1. Relatively few respondents (14.9%; 44/298) reported that they perceived that the quality of their cancer care had been negatively impacted by the Covid pandemic. Relative to respondents who were patients of PM prior to the pandemic, those referred during the pandemic were more likely to report a negative perception (OR: 1.99; 95% CI: 0.90-3.91), although this association was not statistically significant (P = .08; Supplemental Table 1). Similarly, relatively few respondents reported that they felt that having virtual visits had a negative impact on the quality of their cancer care (13.5%). Relative to White respondents, patients in Racialized Groups were approximately twice as likely to report a negative perception of the impact of virtual visits on the quality of their care (OR: 2.21; 95% CI: 0.96-5.08), although this association was not statistically significant (P = .06; Supplemental Table 2). Age was not found to be associated with having a negative perception of virtual care (P = .48).

Figure 1.

Summary of patient responses regarding overall impressions of care.

Discussion

Covid impacted how some cancer care was delivered which has raised concerns regarding potential impact on quality of care. Some practice changes brought about during the pandemic, such as the implementation of virtual care and reduction of use of some laboratory tests and imaging,^14,15 are likely to persist; however, there has been a paucity of literature published to date examining the impacts of the pandemic with a quality lens. Reassuringly, we found that patients rated the quality of the cancer care that they received during Covid as high, and reported mostly positive perceptions of the patient-centeredness of their care. However, the greatest proportion of patients reported negative perceptions of receipt of interpretation services, being informed of changes in care, receiving information about and provision of emotional support, and involvement of family and friends in care decision making. Limitations on in-person visits likely impacted patients’ exposure to advertisements for support groups and psychosocial services available through the cancer center. Additionally, providers may be less perceptive about and responsive to emotional needs when interacting with patients virtually than in-person, or patients may be less likely to interact with other members of the care team, such as nursing or allied health, ²⁹ who may be more likely to focus on their psychosocial needs. Restrictions on essential care partners’ attendance at in-person visits likely limited family and friend involvement in treatment decision making despite the potential benefits of virtual visits for family and caregiver involvement in care. ³⁰ To our knowledge, our study is the first to comprehensively evaluate patient-centeredness of cancer care delivered during the pandemic and the impact of virtual visits on patient perceptions regarding the quality of their care.

Similar to previous work from Rodriguez et al, ³¹ approximately half of the respondents (46.4%) felt that the Covid pandemic triggered some modification to their cancer care; however, relatively few respondents (14.9%) reported that they perceived that the quality of their cancer care had been negatively impacted by it. Keeping with previous reports from our jurisdiction, ³² as well as reports globally, ¹ we found that beyond the shift to virtual visits, changes in the receipt of imaging (28.9%) and surgical procedures (24.1%) were the most commonly reported modifications to care, and that relative to those patients in follow-up care, patients who were newly diagnosed were more than twice as likely to report having experienced a Covid-related change to their treatment (OR: 2.39; 95% CI: 1.21-4.71). In our jurisdiction, those patients already on treatment were prioritized over new starts, ³³ and patients had more difficulty with getting diagnosed through primary care ³⁴ which may have contributed to the greater likelihood of reporting a Covid-related change for newly diagnosed patients.

In contrast to previous reports,^9,10,35 we did not find any unadjusted associations between patient frailty, gender or age, and reporting a Covid-related change in care (P = .63-.75). This likely reflects that the incidence of Covid serious infection in our jurisdiction was comparatively low and maintenance of cancer care was prioritized, ³⁶ so there may have been less need to triage care delivery than in other jurisdictions. Additionally, patient-initiated delays and cancellations were cited as an early driver of Covid-related modifications to care based on findings from clinician surveys. ¹ While our findings agree with prior reports that the most common reason for patient-initiated care modifications was fear of being infected with Covid during an appointment, ¹ of the patients in our study who reported having experienced a modification to their care, only 5.4% had initiated the change.

Consistent with previous reports in breast ³⁷ and prostate cancer, ³⁸ we found that relative to White respondents, patients from Racialized Groups were almost twice as likely to report having experienced a Covid related change to their imaging scans, genetic testing or cancer treatment (OR: 1.91, 95% CI: 1.03-3.54). Additionally, while overall few respondents felt that receipt of virtual visits had a negative impact on the quality of their cancer care (13.5%), relative to White respondents, patients in Racialized Groups were approximately twice as likely to report a negative perception towards virtual visits with regards to the quality of their care (OR: 2.21; 95% CI: 0.96-5.08). These findings are consistent with a previous study which reported that patients from Racialized Groups perceived communication with providers to be poorer in telehealth sessions. ³⁹ Overall, these findings likely reflect pre-existing systemic disparities in access to care, as well as the higher burden of Covid in racialized communities within our jurisdiction. Institutions could take steps to addressing perceived racial biases in virtual care by documenting patient preferences for modality of care in their chart, and updating current diversity and health equity training offerings for staff to specifically address care delivered virtually. Additionally, institutions should partner with patients and caregivers with lived experience to understand how to best address issues of health inequities and bias, when developing virtual care policy and adopting technologies moving forward.

Limitations

The study was undertaken in patients being treated for a gastrointestinal cancer at a single cancer center in Toronto, Canada which may impact the generalizability of the findings to other disease sites or jurisdictions. For questions related to switching oncologist visits to virtual or reporting on reasons for a delay or deferral in care, respondents could only select one response, which may have impacted the findings. For virtual visits we did not delineate between telephone and video as >90% of virtual visits were conducted via telephone at our center at the time of the survey. As virtual visits were primarily completed by telephone at our center, this may impact generalizability as findings for other modes of virtual care delivery may be different. The response rate was relatively low (17%), likely due to respondent fatigue from the number of competing surveys being sent out within the organization and with Covid in general, as well as restrictions on approaching potential participants in-person at the cancer center. As such, the findings are likely influenced by self-selection bias and response bias; and representativeness of the resultant sample may have been impacted. Participation in the survey was restricted to only those patients with a valid email on file which represented 60% to 70% of the gastrointestinal cancer patient population at our center which may have also introduced sampling bias. There was 15.4% to 16.5% missingness across demographic variables in the data set whereby the same set of patients have missing demographics across all of the variables. This was likely due to respondent fatigue as demographic questions were at the end of the survey measure. There were relatively few respondents at either of the extremes of age, and fewer female respondents than males; however, age and gender distributions of respondents were similar to the broader population of patients with a gastrointestinal clinic visit at our center during the same period. Additionally, to examine potential impact of representativeness of the survey sample, iterative proportional weighting was used to generate weight estimates relative to population parameters for ethnicity, language, and education which did not change the direction of associations or the conclusions.

Conclusions

Overall, patients reported relatively high perceived quality of care with the cancer care that they received during Covid. However, greater proportions of patients reported negative perceptions surrounding receipt of interpretation services, emotional support and involvement of family and friends in care decisions making which likely reflects limitations on in-person visits, restrictions on essential care partners’ attendance at the cancer center, and implementation of virtual visits. Patients from Racialized Groups were more likely to report a Covid-related change in their treatment; planning for cancer care delivery beyond the pandemic should consider the impacts of racial biases, particularly in developing policies and procedures around virtual care delivery.