Symptom Distress in Older Adults Following Cancer Surgery

Janet H. Van Cleave, PhD, RN; Brian L. Egleston, PhD; Elizabeth Ercolano, DNSc, RN; Ruth McCorkle, PhD, RN


Cancer Nurs. 2013;36(4):292-300. 

In This Article

Design and Methods


A secondary analysis was conducted using data from 5 nurse-directed randomized clinical trials investigating the effect of care provided by advanced practice nurses (APNs) after surgery. The Vulnerability/Risk/Human Response care model of nursing guided the selection of independent and dependent variables for this study.[18] The model depicts the integrated function of the whole individual from a biopsychosocial view of individual vulnerability (demographics, biologic, psychological) and environmental (treatment) factors that affect human responses (ie, symptom distress).

The study was approved by Yale School of Nursing Human Subjects Institutional Review Board. Informed consent was previously obtained from all patients during the parent studies, and study identification numbers were used in place of personal identifying data in order to protect the confidentiality of human subjects.

Data Source

The 5 clinical studies were conducted between 1983 and 2007 at academic cancer centers in the Northwest and Northeast United States. Complete details of these studies have been provided elsewhere.[19–23] Patients were recruited during their hospitalization, and baseline interviews conducted within 1 month after discharge except for study 1. In study 1, patients were recruited in the outpatient clinics after discharge, and baseline interviews were conducted, on average, 60 days after surgery. The data collection times consisted of baseline (enrollment) and then 3 and 6 months after enrollment. Demographic data and comorbidities were recorded at baseline. Cancer site, stage, and treatment data were collected from treatment records and medical record audits. Symptoms, psychological, and function data were obtained from patient interviews at baseline and at 3 and 6 months. The types of cancers differed among studies. Study 1's population consisted of patients with thoracic cancers. Studies 2 and 3 populations were patients with heterogeneous breast, colorectal, head and neck, thoracic, prostate, gynecologic, bladder, pancreatic, esophageal, renal, and gastric cancers. Study 4 targeted men with prostate cancer. The study 5 participants were women undergoing abdominal surgery for presumed gynecologic cancers; however, the final pathology showed ovarian, uterine, endometrial, and metastatic pancreatic cancers.

The studies were combined to increase the patient population, providing power to detect clinically meaningful changes over time. The procedures used to combine the studies were based on the literature addressing secondary analysis.[24,25] We first mapped the selected variables across the 5 parent studies to examine and plan item compatibility using an Excel spreadsheet. This showed that all studies used similar demographic variables, count of comorbidities, and symptom measure. Multiple scales, though, were used to measure psychological profiles of patients.[26–29] Each scale had 1 consistent measurement item: "blue or downhearted" and thus was used for this study. Studies 1, 2, 3, and 5 used a similar function measure, Enforced Social Dependency Scale (ESDS).[30] Researchers in study 4 used the SF-36 Health Survey (SF-36) physical component summary measures to assess function.[31]

Because of advances in science, pathological reports varied in detail across the studies. For consistency, all pathological reports were analyzed, compared, and updated to the current classification and staging system published in the AJCC Cancer Staging Manual, Sixth Edition[32] and information from the 2008 National Comprehensive Cancer Network Web site ( This review changed the stages of 12 study participants from early to late. These changes were reviewed and verified by a second independent coder.

Participant Selection

A total of 1032 participants were enrolled among the 5 studies, in which 537 were 65 years or older. Eighty-five patients who were 65 years or older died by the 6-month time period. Because of subject attrition, only 428 patients had symptom data collected for a minimum of 2 or more time points. We limited the study population to 326 subjects 65 years or older with a diagnosis in the digestive system, thoracic, gynecologic, and genitourinary sites treated primarily with surgery with no other concurrent malignancy or other medical treatment and minimum of 2 points of symptom assessment data collection. The rationale for limiting the study population to these cancers sites was that the majority of these cancers occur among older adults.[33] Furthermore, the surgery for these cancer sites usually involves an invasive procedure in the thorax, abdomen, or pelvis, influencing older adults' function and symptom distress. Among these 326 patients, 186 patients were listed as undergoing surgical resection of their tumors, and 3 patients were treated with colonoscopies, biopsies, or lumpectomies. The target population for the 5 nurse-directed studies consisted of surgical cancer patients; we therefore assumed the remainder of the surgical data was either missing or not collected.

Main Outcome Measure

The main outcome variable was symptom distress, measured by the Symptom Distress Scale (SDS).[12] The SDS consists of 13 common symptoms of cancer patients: frequency of nausea, severity of nausea, appetite, insomnia, frequency of pain, severity of pain, fatigue, bowel pattern, concentration, appearance, breathing, outlook, and cough. Subjects rated their distress on a scale from 1 (low distress) to 5 (high distress). Total symptom distress is obtained by an unweighted sum of all items, resulting in scores ranging from 13 to 65. The SDS has demonstrated validity and reliability with reported Cronbach α's ranging from .70 to .89.[12,34]


Demographic. The demographic variables for the current study were age, gender, race/ethnicity, marital status, employment, occupation, education, income, and living alone. These data were abstracted from the patient history forms and recorded on a standardized form. All subjects 65 years or older during the 6 months of the study were included, and ages at 6 months were used for the analysis. Age categories typically used in gerontology research were initially used for the present study: 65 to 74 years, 75 to 84 years, and 85 years or older. A review of the univariate statistics of the demographics, however, revealed only 12 subjects (3.7% of population) in age category "age > 85." The categories, therefore, were collapsed to 65 to 69 years, 70 to 74 years, and 75 years or older to provide balance for the statistical analysis.

Biologic. The biologic variables were cancer site, stage, and comorbidities. The cancer site and stage data were obtained from the original study's medical record audits. For the analysis, the cancer types were grouped based on the AJCC Cancer Staging Manual, Sixth Edition classification system[32] and categorized as digestive system (esophagus, gastric, colorectal, pancreatic), thoracic (lung), gynecologic sites (vulva, uterine, ovarian), and genitourinary sites (prostate, renal, renal pelvis and ureter, and bladder). These cancer categories are documented to have increased incidence in older adults.[33] We followed the precedent from original studies and used the same stage categories "early stage" and "late stage." For all cancers except prostate, early stage was defined as stages I and II, and late stage was defined as stages III and IV. In prostate cancer, early stage was defined as I, II, and III, and late stage was defined as stage IV.

Comorbidity was defined as other preexisting health conditions or diseases outside the index cancer. Following procedures from the 5 original studies, all preexisting health conditions or diseases reported by the study participants were categorized as follows: cardiovascular, respiratory, endocrine, eye/ear/nose/throat, mental, neurological, genitourinary, gastrointestinal, skin, cancer, injuries, infectious, and other. The empirical indicator was sum of comorbidities. Evidence from research studies support concurrent and predictive validity for count of disease as a measure of comorbidity.[35]

Psychological. The psychological measures used in the 5 studies were SF-36 Mental Component Summary Scales, SF-12 Mental Component Summary Scales, Profile of Mood States, Mental Health Inventory 5-item version, and Center for Epidemiological Studies–Depression Scale.[26–29] Although multiple scales were used across studies, each had 1 consistent measurement item: "blue or downhearted." The item "blue or downhearted" is reported to be a powerful, nonspecific detector of mental health disorders.[27] Hence, this item was used to represent mental health for the current study.

Treatments. Cancer treatment data were extracted from treatment records and medical record audits from the original studies. All patients in this study underwent surgery as their primary cancer treatment, but may have also received chemotherapy, radiation therapy, or any combination of the 3. Hormone therapy was classified as chemotherapy in the original medical record audits, and this definition was maintained for the present study. Three patients started chemotherapy greater than 3 months after enrollment. None were documented as receiving neoadjuvant chemotherapy. "Nursing intervention" was defined as participation in an oncology APN intervention and operationalized as an assignment of study participants to receive or not receive the intervention. The measurement items for the study were APN intervention versus no APN intervention; surgery alone; surgery and chemotherapy; surgery and radiation therapy; and surgery, chemotherapy, and radiation therapy.

Function. Patients' function is strongly associated with symptoms[36] and hence was included as a covariate in this study. Function was defined as the individuals' actual performance of activities and tasks associated with their current life roles.[37,38] Studies 1, 2, 3, and 5 used a similar function measure, ESDS.[30] Researchers in study 4 used SF-36 physical component summary measures.[31]

The ESDS consists of 2 components: personal and social competence. Personal competence includes 6 activities of eating/feeding, dressing, walking, traveling, bathing, and toileting. Dependency in each activity was reported by the patient and rated by the interviewer on a 6-point scale. Scores for personal competence were summed and ranged from 6 to 36. Social competence consisted of home, work, and recreational activities, which were rated on 4-point scales, and the category communication rated on a 3-point scale. Scores for social competence were summed and ranged from 4 to 15. Scores for personal and social competence were summed to generate a total dependency score ranging from 10 to 51, with higher scores reflecting greater dependency. The ESDS has demonstrated reliability (Cronbach α = .72 to .96) and validity.[30]

The SF-36 is a 36-item survey of health status used to assess 8 health concepts: physical functioning, role limitations due to physical health problems, bodily pain, general health perceptions, vitality, social functioning, role limitations due to emotional problems, and mental health. The SF-36 has demonstrated item-internal consistency and item-discriminant validity. Reliability coefficients have ranged from a low of 0.65 to a high of 0.94 across scales (median = 0.85).[31,39] The measure has also demonstrated validity in discriminating between serious and minor medical conditions as well as psychiatric illnesses.[39]

The mapping of variables described previously demonstrated that the function measures differed among studies: all studies used the ESDS except study 4, which measured function with the SF-36. The 5 studies were judged to consist of compatible study designs and methods. Hence, we combined the 5 studies based on methods described in the literature.[40–42] The data were, therefore, prepared for this study by reverse coding the ESDS scores to attain consistent direction with the SF-36. The data were then converted to standardized norm-based scores (mean, 50 [SD, 10]) using 0–100 metric and z scores. SF-36 scores were converted to norm–based scores using Quality Metric Health Outcomes Scoring Software 2.0 (Quality Metric Inc, Lincoln, Rhode Island).

Statistical Analysis

The outcome variable for the analysis was symptom distress. Descriptive statistics were conducted to determine the mean symptom distress score by age category and cancer site. Multiple linear regression analysis was then used to investigate whether differences in symptom distress across age categories were statistically significant while controlling for demographic, biologic, psychological, treatment, and function covariates. The models were estimated using generalized estimating equations (GEEs) with an unstructured working correlation matrix for greatest flexibility to account for the correlation of observations over time. Dummy indicators were used to study time and interactions between time and age categories to examine time trends on the effects of age categories. The criterion for statistical significance was P < .05.

Sixty-one of the 326 patients did not have complete data; thus, 265 patients were included in the multivariable analysis. An inverse probability of censoring weighted estimator was used because of the rate of attrition and incomplete data over the 3 waves of data. To implement this method, contributions to the standard GEE fitted model were weighted inversely by the probability of not dropping out.[43,44] Such inverse probability weighting accounts for data that are potentially missing at random, which is a less stringent assumption than the missing-completely-at-random assumption of the unweighted GEE fit estimators.