| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
1 Department of General Oncologic Surgery, City of Hope National Medical Center, 1500 E. Duarte Road, Duarte, CA 91010, USA
2 Department of Nursing Research and Education, City of Hope National Medical Center, 1500 E. Duarte Road, Duarte, CA 91010, USA
3 Vital Research, Los Angeles, CA, USA
Correspondence: Address correspondence and reprint requests to: Betty R. Ferrell, RN, PhD; E-mail: bferrell{at}coh.org
Key Words: Palliation • Surgical palliation • Quality of life • Distress • Outcomes
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION METHODSRESULTSDISCUSSIONREFERENCES |
|---|
As this happens, medical practitioners most often define treatment success in terms of survival. However, a patient’s quality of life is increasingly being used as justification in treatment decisions. Medical therapies are now being evaluated and approved using factors that include quality of life. Unfortunately, the surgical literature is devoid of data showing the effect surgical interventions have on symptom distress and quality of life. Although some data exist, few use patient supplied measures to assess the subjective response a given operative therapy imparts.
Despite being understudied, the use of palliative surgery is common.2 However, few, if any, proven indications for surgical palliation exist. With this study, we sought to prospectively determine the impact of palliative surgery on symptom severity, frequency, degree of distress, overall distress, and multidimensional quality of life (physical, psychological, social, and spiritual).
|
METHODS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Data were entered into separate files for each study form, audited, and verified by two expert nurses. Data files were initially managed in SAS (SAS, Cary, NC, USA) then restructured and analyzed using SPSS for Windows Version 12.0 (SPSS, Chicago, IL, USA). Data were audited and corrected for entry errors. Missing data on key static variables were imputed using multiple imputation software. Some variables required a natural log transformation of the data values plus a constant [ln(x+10)] in order to achieve a normal distribution and eliminate 0 values prior to imputation. Following the imputation procedure, log transformed variables were exponentially transformed back to their original ranges and winsorized to eliminate outliers.
The postoperative encounter form and hospital readmission form data sets had multiple records per case and were aggregated for purposes of analysis. The primary variables used in those two data sets were the number of encounters and number of readmissions for each patient. Encounters were defined as any contact, either in person or via telephone, with a physician, physician extender, or supportive care practitioner. In addition, the percent of encounters that were unscheduled, took place in urgent care, or were for new problems were captured. The quality of life and distress thermometer data sets had up to seven records per case due to multiple measurements of the study outcomes over time. These two data sets were each restructured so that there was one record per case, and each record contained all the repeated measurements of each variable uniquely named according to the measurement period. All of the data files were then merged by patient identification number in order to conduct the study analyses. The 2 months time point was chosen for measuring outcomes of the surgery because recovery from the operation was likely complete and the primary tumor was less likely to have advanced, causing other symptoms and confusing postoperative QOL.
Paired t-test were used to examine changes in outcomes between preoperative measures and the 2-month measurement. Specifically, the four QOL subscales, overall QOL, and the single item measures of quality of life were tested, along with the Distress Thermometer score, each of the physical QOL subscale individual items (e.g., pain, nausea), and the rating scales patients used to indicate how frequent or how distressing their symptoms were. Because of inflation of alpha, a Bonferroni correction factor was used to interpret the significance of changes in QOL. Hence, alpha was set at 0.01. For distress, alpha was set at 0.05, and for primary symptom frequency and distress, alpha was set at 0.025 (Fig. 1
).
|
There are several advantages to using growth curve models to assess change. For example, it allows for the inclusion of all 104 patients in the analysis even if they have missing outcome data at several measurement periods. Because all available outcome measurements (compared to a simple pre–post analysis) are included in the analysis, reliability and sensitivity to detecting change are increased. In addition, latent growth curve analysis accounts for error by building measurement error into the model and retains patients’ individual variability by allowing for a range of intercepts, slopes, and quadratic factors across participants. Furthermore, effects can be analyzed by including either time invariant or variant factors.
|
RESULTS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Fifty-three females and 51 males were included in the study. The majority of patients were Caucasian (n = 89; Hispanic = 20; nonHispanic = 69), followed by Asian-Pacific Islanders (n = 13), and African Americans (n = 2). The mean patient age was 55.5 years (range 23–83). Gastrointestinal cancers were the most frequent (48.1%) (Table 1). The mean number of comorbidities per patient was 0.65 (range 0–3). Of the 72 patients presenting with symptoms, the most frequent primary symptom was pain (n = 25, 36%) followed by symptoms of gastrointestinal obstruction (n = 15, 22%) and dyspnea (n = 9, 13%). The mean number of symptoms per patient was 1.30 (range 0–6). Fifty-six patients (53.8%) presented at diagnosis. Twenty-eight patients presented with either stage I or II disease, with the most common being stage III (n = 46, 45.2%). The most common operations were laparotomies (n = 55, 52.9%), laparoscopies (n = 12, 11.5%), and extended skin, soft tissue, and breast resections (n = 8, 7.7%). There were five bronchoscopies, four endoscopies, and three portacath placements.
|
Overall QOL decreased from 3.55 preoperatively to 3.31 at 2 months postoperatively (p = 0.012). At those same time points, physical QOL decreased from 3.77 to 3.37 (p = 0.008), spiritual QOL from 3.81 to 3.57 (p = 0.025), psychological QOL from 3.43 to 3.21 (p = 0.097), and social from 3.10 to 3.00 (p = 0.432). When compared between preoperative and 2-month postoperative, the distress thermometer score decreased from 4.70 to 3.58 (p = 0.032), the frequency of primary symptom decreased from 3.77 to 1.85 (p = <0.001), and the distress caused by the primary symptom decreased from 4.08 to 2.24 (p = <0.001). These are summarized in Table 2.
|
0.001. Prior to surgery, the frequency of primary symptoms is estimated at 3.591 (intercept) and declines by an average of 1.006 (slope) at each time point. However, the positive quadratic parameter indicates that this decline in the frequency of primary symptoms decelerates over time. The variability in the slope and quadratic parameters (p 0.001) is significant. The occurrence of primary symptoms declines until the 5-month follow-up measurement when it begins to increase (Fig. 2).
|
2(19) = 51.835, p = 0.000)], the NFI (0.717) and CFI (0.788) values suggest a moderately good fit of the model to the data. The mean intercept, slope, and quadratic parameters are significantly different from 0 at p 0.001 (Fig. 3). The variability in the intercept (p = 0.002), slope (p 001) and quadratic (p 0.001) parameters are significant. The covariance between the slope and the quadratic parameters is also significant (p 0.001).
|
|
DISCUSSION |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In patients well enough to remain in the study for at least 2 months, the improvements in symptom frequency and distress are marked. Results of the latent growth curve analysis for the entire sample indicate that the gains persist for a few months. However, symptom frequency and distress begin to worsen again at 5-month post-surgery. Specifically, although the decline in the occurrence of symptoms declines markedly immediately following surgery, the decline is also attenuated with time until the occurrence of symptoms is lowest at the 4-month follow-up measurement. At the 5-month follow-up point, the occurrence of primary symptoms starts to increase again. Patients did not differ significantly from each other in the frequency of their symptoms prior to surgery (intercept). However, they did differ significantly from each other in their rate of improvement (slope), suggesting that some patients experienced a steep decline in the occurrence of their primary symptoms while others only experienced a slight decline. The deceleration in the decline (quadratic) also varied significantly between patients indicating that some patients may have maintained their original rate of improvement in symptom frequency while others may have experienced a very quick deceleration in their rate of improvement. These significant values suggest that another variable, rather than measurement error, may be responsible for these differences between patients. The significant covariance between the slope and quadratic parameters suggest that patients who improve at a fast rate also slow down in their rate of improvement much quicker than other patients. These results again highlight the need for future research to examine factors accounting for this variability. The results for primary symptom distress mirror those for primary symptom frequency. However, patients differed significantly from each other in their initial level of distress, indicating that some patients reported high distress prior to surgery while some patients reported low distress. Clearly, the ability to select which patients will benefit from palliative operations will further enhance these positive results and bring about symptom amelioration for a larger number of patients.
Despite improvements in symptom frequency and distress, patients in this sample experienced decreases in overall QOL and many of its subscales. Though statistically significant, these small decreases may not be clinically significant, as the decreases averaged 0.24 on a six-point scale (0–5, inclusive). It is questionable that such a small change in QOL would be clinically apparent in these complicated patients. Moreover, the unyielding nature of the primary tumors would have likely decreased QOL at a quicker rate without intervention. Although further research is necessary to test for differences between patients who receive palliative surgeries and control patients who do not, the results in this manuscript represent striking improvements in symptom frequency and distress with very small changes in QOL.
These results somewhat contradict that from prior studies. Miner and his/her colleagues (2002) found that 12 of 26 patients in their study experiencing an improvement after palliative operation show their improvement around 30 days.3 In our sample, after the initial decrease in QOL at 3 weeks postoperative that is attributed to surgery, the positive effects were seen at 2 months. The results of the growth curve analyses suggest that the improvements in primary symptom frequency and distress will persist beyond the 2-month postoperative period. This difference between studies is likely due to different sample populations with different diseases, as half of their subjects had upper gastrointestinal, pancreatic, and gastric cancers.
Success in palliative medicine is often difficult to define as patients progress in their diseases. While a specific symptom may be ameliorated or eliminated due to either medical, radiologic, or surgical interventions, another may arise as a result of a tumor in another location or from the therapies themselves. To measure a patient’s overall palliation after an intervention, the palliative surgery outcome score (PSOS) was devised.4 The PSOS is defined as the number of symptom-free, nonhospitalized days divided by the number of hospitalized days (up to 180). This measures the proportion of time a patient is out of the hospital and without symptoms and considers the effect of perioperative morbidity. A score of 0.70 was determined by patient interviews to represent a successful therapy.5 The PSOS more easily enables comparisons to be made between palliative interventions and serves as a useful adjunct to QOL and distress thermometer measurements in determining therapeutic success.
In addition to surgeons being reticent to offer operative interventions to patients once there was no longer a possibility of cure, the lack of data proving the utility of surgery in these circumstances slowed the adoption and growth of palliative surgery. At the heart of this was the lack of a clear understanding of the goals and definitions of palliation. This charged with the World Health Organization publication characterizing palliation as "an approach that improves the quality of life of patients and their families facing the problems associated with life threatening illness through the prevention and relief of suffering, by the means of early identification, impeccable assessment, treatment of pain and other problems, including physical and spiritual".6 The Institute of Medicine further clarified this by defining palliation as seeking "to prevent, relieve, reduce, or soothe the symptoms of disease or disorder without affecting a cure".7 Once the field was defined, interventional strategies could be identified and explored.
With well-understood definitions and validated research tools [Functional Assessment of Cancer Therapy (FACT), European Organization for Research and Treatment of Cancer (EORTC)], the use of palliative surgery will increase.8,9 Minimally invasive and endoscopic therapies, often used in this study, will likely continue to expand surgical indications in this difficult patient population. An example of this phenomenon is the use of expandable metallic stents obstructing upper and lower gastrointestinal cancers.10–13 In addition to affecting gastrointestinal continuity, these mechanisms were shown to decrease cost and length of stay while significantly improving QOL. As technology continues to advance, more minimally invasive techniques will become available, decreasing our dependence on open operations. The lower QOL scores seen 3 weeks postoperatively in this study would be reversed, resulting in a more immediate QOL improvement and higher palliative surgery outcome score.
As shown in this study, operations for symptoms in patients in advanced, incurable neoplastic diseases can be successful at decreasing symptom severity and frequency, leading to improvements in QOL and the amount of distress these often very ill patients endure. Patient selection is of paramount importance, as different primary cancers in a variety of locations will likely respond differently to operative interventions. Further prospective study is required to continue to define patient populations that are most likely to benefit from these operations. To do so, surgeons must become advocates to ensure adequate funding mechanisms become available for such studies. It is not until surgeons can determine which patients are likely to benefit from operative palliation that the field will be generally accepted and thrive.
|
FOOTNOTES |
|---|
Received for publication August 8, 2006. Accepted for publication August 9, 2006.
|
REFERENCES |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. D. Smith and L. E. McCahill Predicting Life Expectancy and Symptom Relief Following Surgery for Advanced Malignancy Ann. Surg. Oncol., December 1, 2008; 15(12): 3335 - 3341. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
