«Patterns of sore mouth in outpatients with cancer receiving chemotherapy. By: Carlton G. Brown, Susan L. Beck, Douglas E. Peterson, Deborah B. ...»
Patterns of sore mouth in outpatients with cancer receiving chemotherapy.
By: Carlton G. Brown, Susan L. Beck, Douglas E. Peterson, Deborah B. McGuire, William N.
Dudley and Kathleen H. Mooney
Brown, C.G., S.L. Beck, D.E. Peterson, D.B. McGuire, W.N. Dudley & K H. Mooney. Patterns
of sore mouth in outpatients with cancer receiving chemotherapy. Supportive Care in Cancer,
2009. 17(4): p. 413-428.
Made available courtesy of Springer Verlag. The original publication is available at www.springerlink.com ***Reprinted with permission. No further reproduction is authorized without written permission from Springer Verlag. This version of the document is not the version of record.
Figures and/or pictures may be missing from this format of the document. ***
Goals The aims of this secondary data analysis were to (a) categorize patterns in the development, duration, intensity, and resolution of sore mouth (which can be considered a proxy for oral mucositis) severity and distress over two cycles of chemotherapy in cancer outpatients and (b) examine the relationship of demographic (age, gender, marital status, and educational level) and disease characteristics (type of cancer and type of chemotherapy) to specific patterns of sore mouth (SM).
Materials and methods Visual graphical analysis (VGA) was applied to identify individual patterns of SM severity and distress in 51 outpatients receiving chemotherapy who provided daily reports of sore mouth using a computerized interactive voice response system. The majority were female (n = 41, 8%) with a mean age of 53 (SD = 8.35). Most had breast cancer (68%), and one third received chemotherapy with adriamycin and cyclophosphamide (AC). VGA is a technique in which graphs of individual patients’ symptoms are coded for specific individual or group profiles.
Main results Seven distinct patterns were identified based on variability in onset, duration, and intensity (degree of severity or distress). Chemotherapy agents were significantly associated with patterns of SM. The AC regimen was significantly associated with late onset; however, the intensity did not last long. In contrast, patients receiving R-CHOP were significantly more likely to experience duration intensity (SM after day 15 and a score equal to or greater than a 5 on a 1–10 scale).
Conclusions VGA revealed symptom patterns often hidden in traditional analysis. Understanding individual variability is important to the design and implementation of future intervention research and clinical care.
Keywords: oral mucositis | chemotherapy | outpatient cancer treatments | cancer patients | cancer outpatients | health sciences
Introduction While contemporary cancer therapies can be highly effective in curing or stabilizing the cancer patient, debilitating side effects, such as oral mucositis (OM), continue to compromise delivery of optimal therapeutic regimens. In addition, the toxicities from cancer treatment can affect overall quality of life [12, 15]. Almost 90% of the one million cancer patients treated with chemotherapy in the USA receive these medications on an outpatient basis; thus, OM can be a frequently encountered toxicity in clinical oncology practice . Once patients leave the clinic or physician office, assessment is challenging; thus, OM often goes unreported and untreated.
In spite of an increase in research on OM , there are still important gaps in our knowledge for at least three fundamental reasons. First, most longitudinal research of OM has been conducted with inpatients receiving high-dose chemotherapy for blood and marrow stem cell transplantation [6, 7, 18] or with patients receiving radiation therapy . Few studies have included outpatients receiving chemotherapy as participants; thus, little is known about the trajectories and patterns of OM in this population. In one intervention study, Dodd et al.  investigated the incidence of chemotherapy-induced OM in outpatients over 3 months; however, daily assessments were not completed and results were reported in aggregate.
Second, only a minimal amount of research has been conducted that examines OM longitudinally over an entire course of chemotherapy regardless of whether chemotherapy was administered on either an outpatient or inpatient basis. In addition, few studies have investigated OM over consecutive cycles of outpatient chemotherapy.
Third, most research studies have used standard analytic approaches with summary statistics to explore changes in OM over time; such approaches may mask some of the clinical variability seen in practice. Several important descriptive and intervention studies have included daily measures of OM using a variety of measures. All of theses studies reported change over time using group level summary statistics such as means and standard deviations at each time point [6, 7, 17, 18]. These studies, all conducted in hematopoietic stem cell transplant settings, revealed group level patterns in which OM peaked after myeloablative therapy and then gradually recovered as the bone marrow rebounded. However, the use of traditional statistical methods to summarize group change over time may have obscured individual trajectories of change .
The incidence of OM depends on the particular antineoplastic treatment regimens and patientspecific characteristics. Anthracycline-based regimens are associated with rates of OM around 1– 10%, including those standard regimens for adjuvant treatment in patients with breast cancer (doxorubicin and cyclophosphamide) and regimens for non-Hodgkin’s lymphomas (cyclophosphamide, doxorubicin, vincristine, and prednisone) . Further, there is an increased risk of OM when rituximab is added to the above regimens when used in the treatment of breast cancer . When 5-FU (commonly used in colon and some breast cancer regimens) is administered, the rates of grades 3 and 4 OM are greater than 15% . There is no research to explore whether patterns of OM over time are associated with type of regimen.
There is some limited evidence that gender and age may be risk factors for OM. Vokurka et al.
 reported a higher incidence of chemotherapy-induced OM in females when compared with males. Raber-Durlacher et al.  suggested that patients over 50 years of age develop OM that is more severe and has a longer duration. They proposed that this more intense and prolonged OM might be related to a decline in renal function in patients over 50.
The clinical literature has always consistently suggested a typical pattern of OM in patients receiving chemotherapy. Erythema is typically the earliest clinical sign of OM, occurring approximately 4–5 days following chemotherapy administration . Ulcers then develop 7–10 days following chemotherapy, creating marked discomfort for patients, many of whom require opioid medications for pain relief. OM resulting from chemotherapy administration typically lasts 1 week and usually heals within 21 days after administration . Despite these typical characteristics, few studies examine variation in these trajectories within a given cancer patient cohort, particularly in an outpatient setting.
In summary, there are important limitations to existing research, which include a primary focus on inpatient, high-dose chemotherapy populations, the use of traditional statistical analyses that can sometimes obscure individual trajectories of OM, and a limited number of longitudinal studies investigating OM over more than one cycle of chemotherapy. One key reason for these limitations is the impracticality and cost associated with observational studies of OM when outpatients receive their chemotherapy and return to their home setting. This report takes advantage of an innovative approach to studying symptoms on a daily basis using computerized interactive voice technology. We have specifically focused on daily self-reports of sore mouth (SM) severity and distress, a reasonable proxy for OM used by other researchers [6, 7] that is easy for patients to understand and report.
Thus, the aims of this secondary data analysis were to (a) explore and categorize patterns in the development, duration, intensity, and resolution of SM severity and distress over two cycles of chemotherapy in cancer outpatients and (b) describe and explore the relationship of demographic (age, gender, marital status, and educational level) and disease characteristics (type of cancer and type of chemotherapy) to specific patterns of SM.
Materials and methods Parent study A secondary data analysis was performed using visual graphic analysis to investigate individual patterns of SM severity . The analysis was based on data from a multi-site, prospective randomized clinical trial of patients receiving outpatient chemotherapy  (NCI R01 CA89474).
The purpose of the parent study was to assess the efficacy of telephone-linked care for chemotherapy alerting (TLC-Chemo Alert) in the symptom management of adults receiving chemotherapy for cancer. A total of 223 patients participated in this study.
TLC chemotherapy alerting mechanism was a computer-based communication system that used interactive voice technology to gather and to electronically document cancer patients’ daily symptom experience by means of automated telephone conversations. The TLC-Chemo alert technology is a well-validated technology that included standardized patient response formatting.
The inclusion and exclusion criteria for the parent study sample are summarized in Table 1.
Patients were screened at the end of cycle 1 to confirm that they experienced a symptom that was at least moderate in intensity during their first cycle. Eligible patients who provided informed consent then provided daily data via the TLC system for cycles 2 and 3.
Table 1 Inclusion and exclusion criteria of parent study Inclusion criteria Adult (age 18 or over) Histologic diagnosis of cancer A life expectancy of at least 6 months Cognitively able to participate (as verified by the provider team) At the end of their first cycle of a new course of chemotherapy that was planned for a minimum of 3 cycles Poorly controlled symptoms during the first cycle of the chemotherapy (based on a phone screening) Exclusion criteria Receiving concurrent radiation therapy Biotherapy only Initial demographic and clinical variables were collected from participants at the beginning of the study. Patients were instructed to call into TLC each day during cycles 2 and 3 of chemotherapy and to answer a series of questions that focused on the presence/absence of nine symptoms associated with chemotherapy, including SM. Single-item indicators using a standardized approach for each symptom were utilized—such measures are generally used to obtain the subject’s perception of particular dimensions of multidimensional concepts or of an overall concept. Because data collection was daily, it was essential to keep the response burden to a minimum. Youngblut and Casper  reviewed numerous clinical studies that used singleitem indicators including items to measure symptom severity, concluding that there was evidence to support the reliability and validity of single-item indicators. This approach is also commonly applied in numerous multi-symptom inventories including the Memorial Symptom Assessment Scale and the MD Anderson Brief Symptom Inventory [2, 3].
A hallmark symptom of OM is acute pain of the affected oral mucosa; thus, patients often describe OM as a SM. Use of this term allowed patients to report sore mouth via a pre-recorded script. If patients indicated they had a SM, they were asked specific standardized questions about their SM experience, including how they would rate their level of SM severity and distress on a scale of 1 (minimal) to 10 (worst). A distinct advantage of the dataset is that it provided daily data on the symptom experience through two cycles of chemotherapy.
Visual graphic analysis
Visual graphic analysis (VGA) is a technique in which individual patients’ symptoms are graphed and inspected for patterns over time . VGA provides researchers with an alternative method to quantitative statistical analysis, as it enables careful inspection of individual variations and patterns of change that can be obscured by traditional parametric statistics. A goal of VGA is to disclose common patterns and to pinpoint individuals whose data do not conform to common trajectories. Of particular interest were questions such as when does a SM first occur, how severe does it become, and how long does it last? These questions were best addressed by VGA. The VGA analysis was performed in five steps as described in detail by Brown et al.  and summarized below.
Determining inclusion criteria. The study had two inclusion criteria. First, only those patients who reported three or more episodes of a SM during cycle 2 or cycle 3 of their respective chemotherapy treatments were included. At least three data points are needed for pattern analysis, and reports of sore mouth for more than one or two times would be most consistent with oral mucositis. Second, chemotherapy treatment was limited to the first 21 days of data within each cycle, thus assuring that all graphical plots were based on the same abscissa. For purposes of this study, a cycle was defined as a 21-day period in which chemotherapy was administered on day 1 and not again for another 20 days.
Managing missing data. “No call” missing data occurred when participants never called into the TLC system or terminated a call. These episodes were categorized as missing data and were displayed on the graph as gaps in the plots. Of the 1,554 days available (74 graphs × 21 days), 225 days were missing (14.5%).
Creating individual visual graphs. Data from each episode of SM severity were charted on individual graphs using Microsoft Excel. Identical axes were used for each respective participant or symptom being plotted. This methodology prevented the visual distortion of scales that can occur with many software packages which automatically expand or contract scales to complete a page or plot area .