Daily Patient-Reported Symptoms as Predictors of Unplanned Healthcare Encounters during Chemotherapy: Longitudinal Observational Study

Abstract

Purpose

Chemotherapy can cause symptoms that impair quality of life and lead to unplanned healthcare encounters (UHEs) such as emergency department visits and inpatient admissions. Collecting patient-reported symptom data between clinic visits can enable timely identification of symptoms. This observational study aimed to identify how daily individual symptoms are associated with UHE.

Patients and Methods

We recruited patients (n = 183) receiving cytotoxic chemotherapy for solid tumors into our 90-day study. Participants completed a modified version of the PRO-CTCAE questionnaire assessing common chemotherapy symptoms daily, and UHE information was extracted from medical records. We fit a series of logistic generalized estimating equations to evaluate day-level associations between moderate-severe daily symptoms and the occurrence of UHE within 7 days.

Results

On days with moderate-to-severe symptom levels, the odds of having one or more UHE within 7 days were 4.53 times higher for vomiting (95% CI 1.77, 11.58; P=.002); 2.84 times higher for decreased appetite (1.94, 4.17; P<.001); 2.5 times higher for shortness of breath (1.44, 4.34; P=.001); 1.76 times higher for diarrhea (1.11, 2.79; P=.02); 1.65 times higher for fatigue (1.20, 2.26; P=.002); and 1.58 times higher for pain (1.01, 2.48; P=.04) relative to days with no-to-mild levels of that symptom. After adjusting for days since study enrollment, the odds of UHE were 2.17 times higher on days with moderate-to-severe pain in the abdomen (1.31, 3.58; P=.003). The odds of UHE within 7 days also increased significantly as the number of moderate-to-severe symptoms increased (P<.001).

Conclusions

Specific daily symptoms were associated with increased risk of subsequent UHE during chemotherapy. Interventional studies are needed to better understand if daily remote monitoring of these symptoms can reduce UHE.

INTRODUCTION:

Chemotherapy can cause side effects that significantly impair quality of life¹ and can also lead to unplanned healthcare encounters (UHE) such as emergency department (ED) visits and inpatient admissions² that are costly and distressing to patients³. Over half of ED visits made by patients with cancer are considered potentially preventable⁴, and patient-reported symptom burden is a strong predictor of potentially preventable hospital readmissions in patients with advanced cancer⁵. These findings suggest that earlier detection and management of symptoms may reduce UHE during outpatient cancer treatment.

Severe symptoms including gastrointestinal symptoms and pain are commonly reported by chemotherapy patients at ED visits, but these cases reflect missed opportunities for timely intervention^6–10. Outpatient cancer treatment settings present opportunities for proactive symptom assessment and management to prevent UHE, but physicians tend to underestimate the frequency and severity of symptoms experienced among patients with cancer^11–13. To address these discrepancies, more cancer centers and clinics are incorporating routine patient-reported outcome (PRO) assessment of symptoms at the point of care or prior to scheduled clinic visits^14,15, which may better capture patient symptom experiences. PROs collected at clinic visits have been linked to increased odds of an ED visit or hospitalization within seven days, suggesting that PROs are predictive of UHE^16,17. PROs collected at clinical encounters systematically track symptom burden but fail to capture symptoms between clinical encounters, when chemotherapy side effects typically peak. Collecting more frequent, even daily, PROs between visits may enable earlier identification and management of symptoms before they become severe and escalate into an ED visit or hospitalization¹⁸.

With notable exceptions¹⁸, limited work has investigated associations between daily PROs and UHE during cancer treatment. Daly and colleagues evaluated the value of daily PRO during intravenous antineoplastic therapy, reporting that nearly half of severe symptoms arose without warning and that daily severe symptom alerts increased risk of acute care events, with higher counts of alerts on a given day associated with significantly greater risk¹⁹. While these findings support the clinical value of daily PRO assessment, they did not explore which specific symptoms were most likely to increase risk of UHE. Identifying specific symptoms associated with UHE is essential for delivering efficient care and mitigating obstacles in implementation of PROs such as patient survey fatigue and inundation of alerts to healthcare teams^20,21.

The objective of this observational study is to identify how daily individual patient-reported symptoms are associated with UHE. Our overarching goal is to improve symptom detection strategies, to deliver more timely and proactive symptom management, and to reduce UHE among patients receiving outpatient chemotherapy.

METHODS

Participants

Eligible participants were 18 years or older, were receiving cytotoxic chemotherapy for any solid tumor at UPMC Hillman Cancer Center, an NCI-designated Comprehensive Cancer Center, had at least two chemotherapy cycles remaining, and owned a smartphone. Patients were referred to this study by seven participating oncologists, and the patients enrolled reflect the types of tumors typically treated by these clinicians. The first patient was enrolled in March 2020, just before the University of Pittsburgh halted research activity due to the COVID-19 pandemic. The study protocol was modified to allow participation to be fully remote, and enrollment resumed in August 2020. Once the study resumed, consecutive patients were approached about the study between August 2020 and July 2023. Participants included in the present sample are those who had completed or withdrawn from the 90-day study between March 12, 2020 to October 1, 2023. More information about the study procedure is provided elsewhere²². The Institutional Review Board of the University of Pittsburgh approved all study activities, and informed consent was obtained from each participant.

Measures

Unplanned healthcare encounters

The primary, day-level outcome was the occurrence of one or more UHE within 7 days, including the present day. We defined a UHE as an unscheduled ED visit or inpatient hospitalization. From each participant’s electronic medical record (EMR), author CB extracted the date and reason of any ED visits and the admission date, discharge date, and reason of any hospitalizations that occurred at UPMC institutions during the participant’s study period. Author EK reviewed ED/admission notes to confirm reasons and excluded encounters associated with a planned procedure or device malfunction. If a participant had an ED visit and a hospitalization on the same date, we combined these into a single encounter. We selected a 7-day window based on the prior literature^16,17,19.

For each participant-day, using a fixed 7-calendar day window beginning on that day, a binary label was created by assigning a value of 1 if at least one UHE occurred within the window and 0 otherwise. Due to concern about potential misclassification, days where the 7-day window extended beyond the participant’s study end date and no UHE had occurred during the portion of the window that overlapped with the participant’s study period (i.e., the only portion of the window for which EMR data were available) were labeled as missing.

Daily symptoms

Daily symptoms were assessed via an online survey platform using a modified version of the Patient Reported Outcome-Common Terminology Criteria for Adverse Events (PRO-CTCAE) questionnaire with 24-hour recall²³. Participants were first asked to indicate whether, in the past 24 hours, they had experienced any of the following 16 symptoms common during chemotherapy: anxiety, constipation, decreased appetite, diarrhea, dizziness, fatigue, insomnia, nausea, neuropathy, pain in the abdomen, pain, problems with concentration, problems with memory, sad or unhappy feelings, shortness of breath, and vomiting. For any symptoms the participant endorsed, they were asked to rate the frequency, interference, and/or severity of that symptom using a scale ranging from 0 (never, not at all, or none) to 4 (almost constantly, very much, or very severe). Symptom surveys were completed on patients’ smartphones, which delivered a daily notification reminding participants to complete the online survey at a time of their choice. Patients were informed that their data were not used to trigger clinician alerts, shared with their care team, or used to inform clinical care, and patients were advised to contact their clinical care team with concerns or questions.

Daily survey responses were cleaned by excluding any not associated with a valid participant identifier or less than 50% complete and retaining the first response if participants completed multiple surveys on a given day. For each valid survey response, we computed a series of binary variables reflecting the presence or absence of each of the 16 symptoms at a “moderate-to-severe” level. We first identified the maximum rating across all available attributes (frequency, interference, and/or severity) for each symptom. A value of 1 was then assigned if the maximum rating across all available attributes of the symptom was a 2 or higher and a value of 0 was assigned otherwise, reflecting “no-to-mild” levels of that symptom. Days without a valid survey response were excluded from analysis.

Statistical analysis

We computed descriptive statistics to characterize the sample. To evaluate day-level associations between UHE and daily symptoms, we fit a series of logistic generalized estimating equations²⁴ using the geepack package (v1.3.9) for R²⁵. GEE is a method for modelling clustered data, such as those from a longitudinal study. Either an exchangeable or first-order autoregressive (ar1) working correlation structure was selected for each model by minimizing the Quasi Information Criterion (QIC). Robust standard errors for parameter estimates were obtained using the sandwich estimator. Estimates were exponentiated to obtain odds ratios (OR) and 95% confidence intervals (CI). We used a series of Wald tests to conduct single- and multi-parameter inference. We included in analyses those observations that had both a non-missing value for the outcome and a valid survey response as described above from which daily symptom burden could be characterized (i.e., complete cases).

We first fit a series of univariable models with UHE within 7 days as the outcome and a binary indicator for moderate-to-severe levels of each symptom as the single predictor. We subsequently refit each model adjusting, based on a priori selection, the single covariate reflecting number of days since study enrollment (henceforth “study day”) as a continuous measure of time. For each set of models, we corrected for multiple comparisons by controlling the false discovery rate using the Benjamini-Hochberg procedure. Based on results of individual symptom models, we calculated an additional day-level predictor variable reflecting the number of moderate-to-severe symptoms that were associated with UHE in either unadjusted or covariate-adjusted models at least at trend level (P<.10), to be as inclusive as possible; this variable was binned based on visual inspection of the distribution and examined number of moderate-to-severe symptoms as an additional predictor of UHE within 7 days.

An alpha level of .05 (two-sided) was used as a cutoff for statistical significance. All analyses were performed using R v4.2.3²⁶. Code is available at https://github.com/carissalow/rosa-unplanned-healthcare.

RESULTS

Participants

Between March 2020 and July 2023, 427 patients were approached, of whom 39% (166/427) were uninterested, 18% (78/427) ineligible, and 43% (183/427) enrolled (see Data Supplement, Figure S1). Participant characteristics are presented in Table 1. Participants (n = 183) were aged 59.49 years on average (SD 11.87, range 28–92), 61.7% female (113/183), and 79.8% White/Caucasian (146/183). Most participants had attended at least some college (122/183, 66.7%). Most participants had gastrointestinal (66/183, 36.1%), pancreatic (42/183, 23.0%), or breast cancer (26/183; 14.2%); stage 4 cancer (116/183, 63.4%); and had been diagnosed, on average, 9.98 months before enrolling in the study (SD 20.88, range 0–124).

Table 1.

Participant Characteristics

Characteristic	N = 1831
Age, years	59.49 (11.87) [28, 92]
Gender
Female	113 (61.7%)
Male	69 (37.7%)
Non-binary	1 (0.5%)
Race
Asian	1 (0.5%)
Black/African American	31 (16.9%)
White/Caucasian	146 (79.8%)
More than one race/ethnicity	3 (1.6%)
Other	2 (1.1%)
Highest level of education
Less than a high school diploma	2 (1.1%)
High school diploma or equivalent	39 (21.3%)
Associate of arts or other 2-year degree	18 (9.8%)
Some college, no degree	35 (19.1%)
Bachelor’s degree	48 (26.2%)
Graduate degree	39 (21.3%)
Unknown	2 (1.1%)
Cancer type
Breast	26 (14.2%)
Gastrointestinal	66 (36.1%)
Gynecologic	14 (7.7%)
Pancreas	42 (23.0%)
Urogenital	15 (8.2%)
Other	20 (10.9%)
Cancer stage
0	1 (0.5%)
1	11 (6.0%)
2	26 (14.2%)
3	24 (13.1%)
4	116 (63.4%)
Unknown	5 (2.7%)
Time since cancer diagnosis at enrollment, months	9.98 (20.88) [0, 124]

¹Mean (SD) [Range]; n (%)

Unplanned healthcare encounters

From the EMR, we extracted a total of 98 unique encounters among 60 participants, with 17 direct hospital admissions from outpatient visits (17.3%), 37 ED visits (37.8%), and 44 ED visits with same-day hospital admission (44.9%). Overall, 32.78% of patients experienced a UHE during the study, and the overall incidence of UHE during the study per person-day was 0.62%. Participants with at least one UHE had 1.63 UHE on average during the course of the study (SD 1.07, range 1–6). Characteristics of these encounters are presented in Table 2.

Table 2.

Unplanned Healthcare Encounter Characteristics

Characteristic	N = 981
Encounter type
ED visit and same-day hospitalization	44 (44.9%)
ED visit only	37 (37.8%)
Hospitalization only	17 (17.3%)
Time of encounter, days since study enrollment	39.58 (24.57) [1, 90]
Hospitalization length of stay, days	5.25 (4.49) [1, 20]

¹n (%); Mean (SD) [Range]

A total of 625 days (3.94%) among 60 participants were labelled as one or more UHE within 7 days; 14,146 days (89.16%) were labelled as no UHE within 7 days; and labels were missing for 1,094 days (6.9%).

Daily symptoms

On average, participants completed a daily symptom assessment on 58.4% of enrolled days (SD 28.6%, range 0–100%). A total of 9,510 unique assessments were completed by 178 participants during the study (mean per participant 53.4, SD 25.9, range 1–91); five participants (2.7%) did not complete any assessments and were excluded from analysis. The distributions of the maximum rating across all available attributes for each symptom are displayed in Figure 1. Fatigue (2944/9510, 31.0%), pain (1948/9510, 20.5%), neuropathy (1809/9510, 19.0%), insomnia (1381/9510, 14.5%) and diarrhea (1341/9510, 14.1%) were the symptoms most often classified as moderate-to-severe.

Figure 1.

Distribution of the Maximum Rating Across All Available Attributes (Severity, Frequency, Interference) for Each Daily Patient-Reported Symptom (n=9,510)

Associations with moderate-to-severe individual symptoms

A total of 8,962 observations from 178 participants were included in complete-case analyses. Of those, 321 days (3.58%) among 53 participants (29.78%) were classified as days on which one or more UHE occurred within 7 days. Results of the univariable and covariate-adjusted models characterizing the day-level associations between UHE and the presence of each symptom at a moderate-to-severe level are summarized in Table 3.

Table 3.

Individual Symptom Model Results

Moderate-to-severe symptom		Unadjusted			Study day-adjusted
	N	OR (95% CI)	P value	Q value	OR (95% CI)	P value	Q value
Anxiety
No	8,376	Reference			Reference
Yes	586	1.32 (0.89, 1.98)	.17	.30	1.32 (0.89, 1.95)	.17	.27
Constipation
No	8,430	Reference			Reference
Yes	532	0.82 (0.61, 1.12)	.21	.34	0.82 (0.62, 1.10)	.19	.27
Decreased appetite
No	7,969	Reference			Reference
Yes	993	2.84 (1.94, 4.17)	<.001	<.001	2.77 (1.88, 4.09)	<.001	<.001
Diarrhea
No	7,692	Reference			Reference
Yes	1,270	1.76 (1.11, 2.79)	.02	.049	1.78 (1.13, 2.83)	.01	.04
Dizziness
No	8,622	Reference			Reference
Yes	340	0.98 (0.64, 1.52)	.94	.94	0.98 (0.64, 1.51)	.94	.94
Fatigue
No	6,166	Reference			Reference
Yes	2,796	1.65 (1.20, 2.26)	.002	.008	1.62 (1.19, 2.21)	.002	.008
Insomnia
No	7,661	Reference			Reference
Yes	1,301	0.99 (0.81, 1.20)	.91	.94	1.27 (0.87, 1.85)	.21	.29
Nausea
No	7,854	Reference			Reference
Yes	1,108	1.27 (0.95, 1.70)	.10	.21	1.26 (0.95, 1.67)	.11	.19
Neuropathy
No	7,272	Reference			Reference
Yes	1,690	1.25 (0.96, 1.63)	.10	.21	1.26 (0.97, 1.64)	.08	.17
Pain
No	7,108	Reference			Reference
Yes	1,854	1.58 (1.01, 2.48)	.04	.12	1.56 (1.00, 2.44)	.05	.12
Pain in the abdomen
No	8,001	Reference			Reference
Yes	961	1.21 (0.87, 1.69)	.25	.37	2.17 (1.31, 3.58)	.003	.008
Problems with concentration
No	8,663	Reference			Reference
Yes	299	1.09 (0.75, 1.60)	.65	.80	1.08 (0.75, 1.55)	.69	.85
Problems with memory
No	8,675	Reference			Reference
Yes	287	0.76 (0.33, 1.73)	.51	.68	1.04 (0.53, 2.01)	.91	.94
Sad or unhappy feelings
No	8,411	Reference			Reference
Yes	551	0.96 (0.63, 1.45)	.84	.94	0.96 (0.63, 1.44)	.83	.94
Shortness of breath
No	8,410	Reference			Reference
Yes	552	2.50 (1.44, 4.34)	.001	.008	2.47 (1.39, 4.39)	.002	.008
Vomiting
No	8,852	Reference			Reference
Yes	110	4.53 (1.77, 11.58)	.002	.008	4.52 (1.80, 11.36)	.001	.008

In unadjusted models, there were statistically significant associations between UHE and vomiting, decreased appetite, shortness of breath, diarrhea, fatigue, and pain. Relative to days with no-to-mild levels of the respective symptom, the odds of having one or more UHE within 7 days including the present day were 4.53 times higher on days with vomiting (95% CI 1.77, 11.58; P=.002); 2.84 times higher on days with decreased appetite (95% CI 1.94, 4.17; P<.001); 2.5 times higher on days with shortness of breath (95% CI 1.44, 4.34; P=.001); 1.76 times higher on days with diarrhea (95% CI 1.11–2.79; P=.02); 1.65 times higher on days with fatigue (95% CI 1.20, 2.26; P=.002); and 1.58 times higher on days with pain (95% CI 1.01, 2.48; P=.04) at a moderate-to-severe level. The effect of pain did not survive correction for multiple comparisons (Q=.12).

Results of study day-adjusted models were similar to those of unadjusted models. However, in this set of models, there was a statistically significant effect of moderate-to-severe pain in the abdomen. Controlling for study day, the odds of UHE within 7 days including the present day were 2.17 times higher on days with moderate-to-severe pain in the abdomen relative to days with no-to-mild levels of this symptom (95% CI 1.31, 3.58; P=.003). There was a trend toward an effect of moderate-to-severe neuropathy, but this did not reach statistical significance (adjusted OR 1.26; 95% CI 0.97, 1.64; P=0.08). The effect of the single covariate, study day (days since study enrollment), was not statistically significant in any model (not shown; all P>.11).

There were no statistically significant associations between UHE within 7 days and moderate-to-severe anxiety, constipation, dizziness, insomnia, nausea, problems with concentration, problems with memory, or sad or unhappy feelings in either unadjusted or study day-adjusted analyses (all P>.10).

Associations with number of moderate-to-severe “significant” symptoms

Eight of 16 symptoms were found to be associated with UHE within 7 days at least at trend level (P<0.1) in individual symptom analyses, including decreased appetite, diarrhea, fatigue, neuropathy, pain, pain in the abdomen, shortness of breath, and vomiting. For each day for each participant, we computed the number of these “significant” symptoms which they endorsed at a moderate-to-severe level (possible range 0–8). Because the distribution was right-skewed, this variable was binned into the following categories: 0, 1, 2, 3, or 4 or more (4+) symptoms.

In the unadjusted model, there was a statistically significant overall effect of number of moderate-to-severe “significant” symptoms (P<.001), and the odds of UHE increased as the number of symptoms increased (Table 4). Relative to days with no such symptoms, the odds of the occurrence of one or more UHE within 7 days including the present day were 1.19 times higher on days with 1 symptom (95% CI 0.79, 1.79; P=.40); 1.97 times higher on days with 2 symptoms (95% CI 1.17, 3.32; P=.01); 2.13 times higher on days with 3 symptoms (95% CI 1.33, 3.39; P=.002); and 5.32 times higher on days with 4 or more symptoms (95% CI 2.91, 9.72; P<.001). Results of the study day-adjusted model were similar. The effect of study day was not statistically significant (not shown; P=.14).

Table 4.

Number of Significant Individual Symptoms Model Results

		Unadjusted		Study day-adjusted
	N	OR (95% CI)	P value	OR (95% CI)	P value
Number of symptoms			<.001		<.001
0	4,051	Reference		Reference
1	1,926	1.19 (0.79, 1.79)	.40	1.19 (0.80, 1.77)	.40
2	1,503	1.97 (1.17, 3.32)	.01	1.93 (1.15, 3.24)	.01
3	895	2.13 (1.33, 3.39)	.002	2.08 (1.31, 3.28)	.002
4+	587	5.32 (2.91, 9.72)	<.001	5.19 (2.86, 9.41)	<.001

Figure 2 shows example daily symptom profiles among four participants with and four without UHEs during the study (depicted with asterisks and black vertical lines), including the highest attribute ratings and number of symptoms during the seven-day window prior to each UHE.

Figure 2.

Example daily symptom profiles among (A) participants who had one or more unplanned healthcare encounters (UHE) during the study and (B) participants who did not. The number of days since study enrollment is plotted along the x-axis and individual symptoms are plotted along the y-axis; symptoms that contributed to the number of “significant” moderate-to-severe symptom outcome are displayed in blue text. The green-red color scale reflects the maximum rating across all available attributes for the symptom, with green corresponding to 0 (or not endorsed) and red to 4. The white-blue color scale reflects the number of “significant” symptoms present at a moderate-to-severe level, with white corresponding to 0 and the darkest blue to 4 or more. Missing values are displayed in grey. For participants who had a UHE, the day of the UHE is indicated by an asterisk and the 7-day window surrounding the UHE is indicated by black vertical lines.

DISCUSSION:

This observational study links specific daily smartphone-reported symptoms to the risk of subsequent UHE during chemotherapy. UHEs were common, with 33% of patients experiencing at least one ED visit or unplanned hospitalization during this 90-day study. Our results indicate statistically significant associations between UHE within 7 days and moderate-severe vomiting, decreased appetite, shortness of breath, diarrhea, fatigue, pain, and abdominal pain in a sample of patients receiving chemotherapy for solid tumors. However, no significant associations were found between UHE and moderate-severe constipation, dizziness, insomnia, nausea, problems with concentration or memory, or sad or unhappy feelings. A significant dose-response relationship was observed between number of moderate-to-severe “significant” symptoms and UHE, with fivefold increased risk associated with days on which four or more significant symptoms were endorsed.

These findings are largely consistent with existing literature on high-risk symptoms reported by patients with cancer in the ED as well as via PROs captured at the point of care^2,16. These findings are also consistent with the results of Daly et al.’s study that found that pain and shortness of breath were the most common symptoms to generate alerts and UHE risk increases as the count of severe symptoms accumulates¹⁹. Our findings suggest that daily assessment of the seven specific symptoms linked to UHE risk may be particularly important for early detection of symptoms likely to lead to ED visits or hospitalization.

A strength of this study is daily systematic assessment of 16 symptoms common during chemotherapy, including physical side effects (e.g., neuropathy) as well as psychological symptoms (e.g., sadness). This study is not without limitations. First, this study is from a single institution which might not reflect this patient population at large. Our models did not account for whether daily symptoms were persistent or progressing, and consideration of these temporal symptom patterns might better capture clinically concerning symptoms. Patient surveys were not completed every day²² and patients may have been less likely to complete surveys on high symptom burden (when feeling acutely ill) or low symptom burden (when symptom ratings seemed less relevant) days. Of note, patient surveys collected during this study were not used to trigger clinician alerts or to inform clinical care, and participants may be more motivated to engage with daily PROs if they know this information is guiding clinical management. In addition, given our focus on identifying temporal patient-reported predictors of UHE risk, our analyses did not adjust for demographic or clinical covariates that might be related to overall UHE risk. Likewise, this study did not adjust for days since last chemotherapy treatment given heterogeneity of cancer types and treatments in our sample as well as missing data about chemotherapy schedules, although this will be important for future work to consider. Finally, recruitment during this study occurred during COVID-19 which might have influenced participants’ healthcare utilization behavior.

These findings should be confirmed in a larger interventional study to better understand the clinical utility of daily individual symptom monitoring and its potential to reduce preventable UHE. A number of systems have been developed to collect PROs at or between visits and generate automated alerts to clinicians when severe symptoms are reported for real-time symptom management^{18–20,27–31}. This work highlights the potential of daily symptom monitoring to enable earlier clinical management of symptoms during chemotherapy¹⁸ and to reduce UHE^19,28.

CONCLUSION

This study identified seven symptoms that were significantly associated with elevated risk of UHE, with increased UHE risk showing a dose-response relationship with multiple moderate-severe symptoms. We are in the preliminary stages of optimizing remote monitoring of daily symptoms to promote timely toxicity management and prevention of UHE.

CONTEXT SUMMARY.

Key Objective

The objective is to identify how daily patient-reported individual symptoms are associated with unplanned healthcare encounters (UHE) for patients with cancer receiving chemotherapy.

Knowledge Generated

There were significant associations between UHE within 7 days and moderate-severe vomiting, decreased appetite, shortness of breath, diarrhea, fatigue, pain, and abdominal pain. A significant dose-response relationship was observed between number of moderate-to-severe symptoms and odds of UHE.

Relevance

UHEs are common complications for patients with cancer receiving chemotherapy. The seven significant symptoms identified in this study should be closely monitored to promote real-time toxicity management and prevent UHE.