The study was conducted at three managed-care organizations that had integrated delivery systems and computerized cancer registries. We included all women aged 18 years and older who died of epithelial ovarian cancer during the six-year period 1995–2000. This study period precedes the establishment of palliative care programs at the health plans, with care having been provided primarily by the oncologist and hospital staff. At one plan, only women diagnosed in 1995 or later were included, whereas at the other two plans, women may have been diagnosed earlier than 1995. Enrollees who died from ovarian cancer were identified by matching enrollment databases at the health plans to data from cancer registries and information on deaths from vital statistics offices at the State level. Each case's cause of death was confirmed by abstracting the chart, and women whose primary cause of death was not ovarian cancer or a complication of that disease were excluded (n=13). In addition, women whose charts were missing or substantially empty were not included in the study (n=16). In all, 421 women were eligible for the study.
Two methods were used to collect data: a) linkage with electronic data, and b) manual review of medical records. Data pertinent to this report included the patient's demographic factors (age, race/ethnicity, marital status, location of residence); tumor characteristics; recurrences; information about signs and symptoms at the end of life, including interventions; and comorbid conditions. Most information was obtained for the last six months of life (or from diagnosis to death for women who died within six months of diagnosis).
Linkage with Electronic Data
Variables obtained from linkage included information on the diagnosis and the death, the concentration of cancer antigen 125 (CA-125), use of psychotropic drugs, and comorbidity. In general, data were extracted from computerized information systems for the six months before death; data on comorbidity were based on the period 18 months to six months before death. Information on CA-125 was obtained from computerized laboratory tests, whereas information on dispensing of psychotropic medications was obtained from outpatient pharmacy records.
Review of Medical Records
Variables obtained from review of the medical record included information on the patient's demographic and social characteristics; complications, recurrences, and medical procedures; and patient preferences for stopping curative care and accepting the end of life. Interventions were ascertained for five major complications: ascites, pleural effusion, bowel obstruction, bladder obstruction, and nutritional disorders. Where there was no record of treatment, we assumed that treatment had not been provided. Trained medical record analysts were provided with a standardized instrument and codebook for abstracting the chart. The codebook included question-by-question specifications and background materials as well as abstracting protocols. Direct input of data was achieved through a computerized data entry and management program developed using Microsoft Access® that included range and logic checks. To further ensure quality and consistency of the data across the three sites, we held biweekly conference calls involving all study personnel, including abstractors, to answer questions and review their productivity and the quality of work.
We examined for specific time periods before death the proportion of women who experienced certain signs or symptoms, underwent various measurements, and received procedures of interest. Logistic regression was used to model the association of age at death, race or ethnicity, marital status, stage of disease, interval between diagnosis and death, and comorbidity with the probability of receiving an intervention for complications incurred during the last six months of life. Because the literature on palliative care does not consider declining appetite or thirst to be a reason for intervening with the dying patient, we did not treat nutrition-related complications as representing poor outcomes in this study. We could not consider interventions for fatigue/weakness, nausea/vomiting, constipation, edema of extremities, or anemia, as we did not obtain data on how these problems were treated.
About 90% of women were aged 50 years or older at diagnosis and death (Table 1
), all but 17% were white non-Hispanic, and 50% were married or living with a partner. Almost one-quarter of the women lived alone, 2% in a nursing home, and most of the remainder in a home with other people. Of those with unknown place of death, 51% were referred to hospice, 24% died in the hospital, 9% in a skilled nursing facility, and 36% in a private home. About three-quarters had been diagnosed initially with metastatic disease. The average CA-125 concentration during the last six months of life varied considerably. The interval between diagnosis and death was less than one month for 11% of patients, but three years or more for 14%. Just over half of the patients had no comorbid conditions, 33% had one or two, and 14% had three or more.
Table 1Characteristics of the 421 Study Participants
The most common complications recorded in the medical record during the last six months of life were fatigue or weakness (75%), nausea or vomiting (71%), constipation (49%), edema of the extremities (44%), and anemia (34%) (Table 2
). Other than nutrition, the most common major complications were ascites (28%), bowel obstruction (12%), pleural effusion (10%), and bladder obstruction (3%). Disorders of nutrition were indicated in the medical record with a range of descriptors, from “loss of appetite” to intensive interventions such as parenteral nutrition (9%).
Table 2Complication and Concomitant Intervention Occurrence in the Six Months Prior to Death
About two-thirds of women with ascites were not treated for the condition. Pleural effusion was treated in about half of the women with this problem. For bowel or bladder obstruction, there was no treatment in about one-third of affected women. Four women had abdominal surgery to treat effusion or a bowel obstruction during the last 14 days of life. Thirty-six percent of women had a tumor recurrence during the last six months of life. Thirty-five percent of women received antianxiety medications during the last six months; only 7% used antidepressants during that time, but that number rose to 10% for the final 14 days.
Nutrition-related interventions declined noticeably with the approach to death (Table 2
). Dietary supplements were given fairly often to women who experienced a nutritional problem in the final six months of life, but they were infrequently supplied in the final 14 days (1.2%). Parenteral nutrition support was somewhat less common than dietary supplements during the final six months, and in the final 14 days, very few women received this intervention (1.4%). Tube feedings were infrequent (less than 2% through the final six-month period).
Differences were observed by age at death for the probability of receiving an intervention for pleural effusion, ascites, or obstruction of the bowel or bladder. After adjustment for marital status, race or ethnicity, comorbidity, and the interval between diagnosis and death, women who died at younger ages were more likely than older women to receive an intervention (data not shown), with the odds ratio for each decade of age equaling 0.71 (95% confidence interval=0.53, 0.94, P for trend=0.02). No differences were found in the likelihood of intervention by race or ethnicity, marital status, stage of disease, interval between diagnosis and death, or comorbidity.
In this investigation of women who died of ovarian cancer, we observed that for some complications that might lead to a relatively difficult death, there was no record of treatment, and we concluded that treatment was not offered. Failure to offer treatment was seen frequently for women with pleural effusion, ascites, and obstruction of the bowel or bladder. In addition, numerous women with many months of life remaining were not offered treatment for a major complication. In the case of pleural effusion, we found no record of intervention for 45% of the women who suffered this problem in the last six months of life. The treatment for pleural effusion, thoracentesis, poses a slight risk for infection and damage to the lung itself, but the dyspnea and chest pain that result from this complication can frighten patients and their caregivers. Because the ultimate goal of palliative care is to maximize the patient's comfort, the relief that treatment could provide may well compensate for the risk of the procedure.
Obstruction of the bowel and bladder are most often treated with abdominal surgery. The potential benefit in life expectancy from surgery can be as long as several months. However, this must be weighed against the operative risks, especially among women with poor health status.
We observed a sharp increase in compromised nutrition, together with a sharp decline in management, during the last two weeks of life. Early on in the dying process, maintenance of nutrition may be used to achieve specific ends, such as to give the woman enough stamina to participate in a special event. When death becomes very close, there may not be hunger, and maintaining nutrition may no longer be relevant. In addition, feeding tubes and parenteral nutrition add to the risk of infection, may be poorly tolerated in these patients, and may cause cramps and diarrhea, thereby reducing quality of life.
The literature on palliative care is clear that hydration, per se, is not a goal of treatment in the dying patient.
Dehydration among terminally ill patients: an integrated ethical and practical approach for caregivers.
Do dying patients really need IV fluids?.
Benefits of dehydration in terminally ill patients: allowing terminally ill patients to dehydrate is an ethically and legally acceptable therapeutic measure.
Indeed, it has been proposed that dehydration in dying patients has some therapeutic value (e.g., reduced incontinence, less nausea and vomiting, and fewer pulmonary secretions) and may even produce some analgesic effects. Furthermore, interventions for compromised nutrition, such as the placement of feeding tubes and implementation of parenteral nutrition, pose risks for infection.
Two limitations of our study should be borne in mind. First, a retrospective examination of the final six months of life involves a different perspective from that of the physicians and other professionals who cared for the women in this study. In the actual care environment, for example, the perceived terminal phase could have started well before the final six months for some women, and treatment for some of the problems we examined may have actually been given prior to our study period (while we classified it as not given because it was not in the final six months). The actual effect of these limitations on our results would probably have been modest, however.
The second limitation concerns the use of medical record data for ascertaining complications and their treatment. Use of medical records for ascertaining symptoms among dying patients was examined in a study by Stromgren et al.
- Stromgren A.S.
- Groenvold M.
- Pedersen L.
- et al.
Does the medical record cover the symptoms experienced by cancer patients receiving palliative care? A comparison of the record and patient self-rating.
of Danish patients with advanced cancer. These investigators assessed the concordance of the medical record with patients' self-reported symptoms and found the medical record to provide lower estimates of symptoms than those provided by patient reports: fatigue (only 38% of patients' reports were noted in the medical record), nausea/vomiting (52%), constipation (38%), depression (25%), anxiety (20%), and reduced appetite (51%). Their study included a range of cancer patients, but they did not report on the major complications of interest in the present study (ascites, pleural effusion, obstruction of bowel or bladder, and cachexia). Similarly, use of oxygen for dyspnea and other comfort measures could not be reliably obtained.
Our findings are broadly consistent with a prospective study of symptoms among 1,000 patients with any type of cancer admitted to the Palliative Medicine Program of the Cleveland Clinic.
- Walsh D.
- Donnelly S.
- Rybicki L.
The symptoms of advanced cancer: relationship to age, gender, and performance status in 1,000 patients.
The most prevalent symptoms observed, other than pain, were fatigue, weakness, anorexia, lack of energy, dry mouth, constipation, early satiety, dyspnea, and weight loss, with these symptoms occurring in 50%–84% of patients. One challenge with comparing results across studies is the imprecise definition of these symptoms and lack of standardized assessment. Standardized assessments have been tested in Germany, with large variation in documentation among palliative care units across the country.
- Radbruch L.
- Nauck F.
- Fuchs M.
- et al.
What is palliative care in Germany? Results from a representative survey.
These authors recommend, and we concur, that standardized instruments are the first step toward furthering palliative care research, with training to standardize definitions following close behind.
Another consideration may affect the generalizability of the study: it was set in three managed-care settings where patients have comprehensive health insurance coverage and do not face the overwhelming financial barriers to access that challenge uninsured patients. Hence, the observed variations in patterns of care for women with terminal ovarian cancer are more likely to be associated with the preferences of the patient, family, or provider, and with the managed-care structure of the delivery system.
Besides seeking to identify opportunities to improve palliative care for women dying of ovarian cancer, we hoped to provide information that might be shared with patients and families with which to create expectations about their health status during the final months of life. In a study by Doyle et al.,
- Doyle C.
- Crump M.
- Pintilie M.
- Oza A.M.
Does palliative chemotherapy palliate? Evaluation of expectations, outcomes, and costs in women receiving chemotherapy for advanced ovarian cancer.
27) of women with ovarian cancer who received palliative chemotherapy thought the treatment would have a moderate-to-high chance of curing their disease. This raises concerns that women with ovarian cancer misunderstand the objectives of end-of-life care and that they have unrealistic expectations about their chances for cure, which may lead them to demand aggressive care that reduces the quality of their life before death. An example would be abdominal surgery, although we observed only a small number (n
4) of abdominal surgeries undertaken during the last 14 days of life and cannot comment on their appropriateness.
Since this study was begun, numerous palliative care programs have been established. We recommend evaluation of those programs using prospective study designs with systematic assessment of patient-reported pain and physical and psychological functioning. The perspectives of patients and caregivers and the role of functional status and cultural differences should be evaluated as well.