Therapy-Related Symptom Checklist Use During Treatments at a Cancer Center

Phoebe D. Williams, PhD, RN, FAAN; Kathleen M. Graham, MS, BSN, RN, APNP-BC, AOCN, CNS; Deborah L. Storlie, RN, NC; Therese M. Pedace, BSN, RN; Kurt V. Haeflinger, BS; David D. Williams, MPH; Diane Otte, MS, RN, OCN; Jeff A. Sloan, PhD; Arthur R. Williams, PhD, MA, MPA


Cancer Nurs. 2013;36(3):245-254. 

In This Article


Study Design

This is a quasi-experimental sequential cohort design with cyclical turnover.[41–44] This design is adequate to assess pre-post changes between control and intervention groups, if there are minimal confounding historical changes, adequate statistical controls, and no changes in personnel or other treatment resources. In this study, G1 and G2 were treated within relatively narrow time frames mitigating time-related confounding. Changes in the primary outcome (HRQOL) were adjusted using patient baseline measures and covariates discussed in the literature.[45] Total number of staff full-time equivalents remained constant throughout the study periods. The practice model also remained unchanged throughout the study periods. The study was approved by the study site's institutional review board.

Participants, Sample, and Setting

Randomization of patient assignment into treatment cohorts was impossible in this study, because 1 outpatient clinic served the site and the intervention would itself alter the delivery of care within the clinic. Use of the TRSC by patients and clinicians would lead to changes in interactions between them and among clinicians. Data, therefore, were collected from 2 cohorts of patients sequentially: one cohort before use of the TRSC (G1) at the clinic and another during TRSC use (G2). Study patients were accrued as they were usually scheduled for treatment at the clinic during each observation period. Eligible patients were invited to participate in the study and an institutional review board–approved informed consent was obtained. Inclusion criteria required that study patients had at least 1 day of treatment (radiation, chemotherapy, or both), were not already participating in an ongoing clinical trial, had no diagnosed psychopathology, were 18 years or older, spoke/read English because study forms were not available in other languages, and had an Eastern Cooperative Oncology Group score of 3 or less or Karnofsky score of 60 or greater.

The 2 study cohorts were accrued and followed to the end of each study period: first G1 patients (March 6, 2007, through November 23, 2007) and later G2 patients (May 7, 2008, through March 3, 2009). Repeated measures were obtained from 55 patients in G1 and 58 patients in G2 up to and including a predesignated last date of data accrual for each cohort. As a result of the accrual process, the number of patient visits or observations ranged from 1 to 11. Two patients with only a single visit (1 each in G1 and G2) are excluded from analysis and discussion in this article. There was no overlap between patients in groups 1 and 2.

Sample Size

Equations in Twisk[45] were used to estimate power and sample size for the primary study outcome, the HRQOL-LASA total score. An effect size = 0.30, [alpha] = .05, 5 repeated measures per subject, and a desired power = 0.80 led to 55 subjects needing to be obtained from each of the 2 study cohorts (groups) or 110 subjects in total. To allow for withdrawals, losses to follow-up, and incomplete data collection, 64 subjects were actually accrued into each study group.

Demographics of the 113 subjects used in this article appear in Table 1 . Of the 128 subjects on whom data were collected, 2 were excluded for having only 1 visit, and 13 were not able to be staged for their cancer. Stage was considered a critical covariate in the 2 GEEs. Exclusion of these subjects left 113 for the analysis in this article with 696 observations. The mean number of observations per patient was 5.2.


The study oncology clinic is located in a small city in the US upper Midwest, with population of 107,120 (2005), of which 96%+ is white. The county has 7 major manufacturing companies, 1 state university, 1 private university, 1 technical college, 2 integrated healthcare delivery systems, and other employers in smaller manufacturing companies. The median income of residents was $26,030 (75% of the state's median income). The other 10 counties in the tristate catchment area are similarly situated in terms of mix of economic activities with median incomes below medians in Wisconsin, Minnesota, and Iowa.

The clinic is part of 1 integrated delivery system with 3 hospitals (total of 430+ beds) serving residents in the states mentioned. Cancer services include inpatient and outpatient care at a state-of-the-art Center for Advanced Medicine and Surgery, which opened in 2004 in the city. The physicians in the Division of Hematology/Oncology and Division of Radiation Oncology are all specialists in cancer care. Medical school, residency, and fellowship experiences are from top schools in the United States; all are board certified in their respective specialties, including internal medicine, medical oncology, hematology, and radiation oncology. The nursing staff consists of highly qualified nurses, the majority of whom are oncology certified. Oncology patients also have access to a full set of support services at the clinic.

Standard of Care

Treatment options in the study clinic include chemotherapy and radiation therapy. A wide array of support services is available including assessment and referral for specialized individual and family counseling, support groups, education and resource materials, referral to complementary therapies by patient request, nutritional counseling, social work services, and financial assistance. Documentation of symptoms and their management is done by physicians and nurses using the standard clinic interview and medical record. Health-related quality of life is not documented except when a patient is enrolled in an RCT that requires this. Randomized clinical trials have been exclusively drug trials.

Upon admission for treatment and outpatient follow-up, patients have the option of meeting with a personal cancer guide, an experienced registered nurse. The RN "Cancer Guide" helps the patient and a family member if present develop an individualized plan of care that encompasses psychosocial, emotional, and spiritual needs. Nationally recommended research-based clinical protocols are used for all cancer treatments.


The first study cohort received standard care at the clinic. Patients were entered into the study during the first 4 months and followed up for a subsequent 4 months. Data were collected from these patients using the HRQOL-LASA, a short demographic data form, a symptom management record and Karnofsky scale, and a medical record review by an experienced RN.[46] Patients who received RT daily completed instruments once weekly on the same day each week. Patients receiving chemotherapy completed the instruments on the day of provider evaluation prior to receiving chemotherapy on day 1 of each cycle. The number of radiation treatments and chemotherapy cycles varied, depending on the treatment protocol. The study coordinator monitored and logged in subjects' schedules.

The second study cohort received the standard of care, and the same instruments were used. In addition, the 25-symptom TRSC was completed by patients prior to clinical consultation.[4–6] The checklist was answered on a scale of 0 (none or no symptom) to 4 (very severe). The completed TRSC was provided to the clinicians for use or reference during the patient appointment.

Prior to the beginning of the study, clinic staff were trained in the use of the study instruments and the importance of complete and consistent follow-up to accrue at least 5 complete sets of instruments from each patient. Before the beginning of accrual of the second or intervention study cohort, clinic staff was introduced to the TRSC and advised that on the patients' completion of the form the provider was to be given a copy. No training was provided as to how this form was to be used with patients. Patients were to be accrued as usual by the clinic.

Study Variables and Instruments

The study variables are shown in Chart 1, and instruments or measures used, including the symptom documentation for the non-TRSC group, in Chart 2. During G1, a trained oncology nurse abstracted the symptoms managed and documented after each clinic visit; another nurse checked accuracy. These data were then entered as string variables into a computer database. On a regular basis, these data were checked by the data analyst, the principal investigators and consultants to the study, and members of the research team.

All instruments used have good psychometric properties, including the TRSC.[4–9,27] The major outcome measure is the HRQOL-LASA.[37,38] The TRSC is shown in the Appendix.


Therapy–Related Symptoms Checklist (TRSC)

Data Analysis

A statistician not involved in the collection and review of data conducted the statistical analysis. Stata version 11.0 was used for all analyses.[49] The Stata routines are adjusted for censoring, and semirobust SEs were obtained. Only subjects with 2 or more clinic visits and documented stage were retained in the analysis.

The first study hypothesis was examined using the HRQOL measures in a panel analysis: specifically, a population averaged GEE with a Gaussian distribution, identity link, and exchangeable correlation matrix. Health-related quality of life was measured during visits 1 to 11. The first observation of HRQOL was used as a baseline covariate in the GEE. Because of the use of an HRQOL baseline, the final analysis was done with 583 observations of 113 subjects. The key study variable was group placement. Covariates in addition to the baseline HRQOL were education level attained by the respondent, age in years, gender, whether a significant other is in household, stage, radiotherapy, chemotherapy, and by implication combined treatment if radiotherapy = 1 and chemotherapy = 1. Time measured as days postbaseline was also included as a covariate. This equation was tested for interaction effects among the independent variables, and none were found.

The second study hypothesis was examined using a similar GEE; however, the dependent variable was total number of reported symptoms at each clinic visit (1–11 [mean, 5.2 visits per patient]). This analysis was done with 113 subjects and 696 observations. Covariates were those mentioned previously plus an interaction effect measured as days from baseline multiplied by group placement: this variable is used to capture the combined effects of time and group placement. The multiplication of independent variables in a regression equation is commonly used to capture the effects of 2 or more variables acting together on a dependent variable, thus the term interaction effects.[47]