|Home | About | Journals | Submit | Contact Us | Français|
Objectives To see how often families in Oregon reported moderate to severe pain in dying patients in late 1998 compared with late 1997. Design A systematic random sample of death certificates was used to identify family members of decedents who died in a hospital setting between October 1 and December 31, 1998. A structured telephone interview was used to obtain data. Participants Family members of 103 decedents (who died in hospitals 2 to 4 months before data collection) were identified with the use of death certificates. Results In late 1998, 56 family members (54%) reported that their loved one experienced moderate to severe pain in the last week of life. Conclusions Family reports of moderate to severe pain in dying hospitalized Oregonians remain high. The influence of environmental factors on pain management may have implications for practice and policy nationwide.
The undertreatment of pain in the dying has been documented1,2 and is a national concern.3 Despite the widespread availability and adoption of pain management guidelines,4,5 numerous investigations have shown that pain remains a significant problem in end-of-life care.2,6 For example, the SUPPORT investigators reported that half of family members of patients who were conscious before death said the decedent had moderate to severe pain most or all of the time.1 Inadequacies in pain management have been part of the impetus for the current national wave of efforts to improve care of the dying, with funding support by numerous foundations and federal agencies and educational programs designed by professional organizations such as the American Medical Association and the American Nurses Association.
In a previous interview study with 475 recently bereaved family members, we asked family respondents to rate their loved one's level of pain in the last week of life on a 4-point single-item indicator (none, mild, moderate, and severe).7 Comparisons by location of death showed that rates of moderate to severe pain were higher for hospital deaths. On closer inspection, however, differences between locations in reports of moderate to severe pain for the first 11 months of the study were not significant. The observed difference by location was due to increased family reports of moderate to severe pain for hospital deaths only during the last 3 months of 1997, which increased from 33% to 57%.7,8
We wondered whether the increased number of reports of moderate to severe pain in dying hospitalized patients noted in late 1997 had returned to baseline rates by late 1998. To address this question, we collected family data about hospitalized decedents who died during the final 3 months of 1998. We were specifically interested in family reports of pain in dying hospitalized decedents and did not anticipate that any other variables would change significantly.
Methods were identical to those used in the initial 1996-1997 study but modified to exclude deaths in private homes and nursing homes.7 The design of the study of human subjects was approved by the University Institutional Review Board and the Oregon Health Division.
In collaboration with the Oregon Health Division, we conducted a systematic random sampling of Oregon death certificates for deaths occurring in hospitals in the last 3 months of 1998 (n = 401). Because Oregon death certificates list the name of an informant (usually a family member) but no address or telephone number, we had to search publicly available records (for example, state telephone directories, newspaper obituaries) to locate family contacts. This proved to be both time-consuming and difficult. People with common last names or unlisted telephones were especially difficult to locate. Using these methods, we located 51% (n = 203) of family contacts. “Family” was defined broadly to include partners, friends, and other people closely involved with the decedents' death because not all those listed as an “informant” on the death certificate were legally family.
In an initial telephone call, the study was explained, and if the family contact agreed, a telephone interview appointment was scheduled for the following week. Despite the challenges of making cold telephone calls, 51% (n = 103) of the family contacts located agreed to participate and were qualified to be included in this sample, thus yielding an overall response rate of 26% of the original sample. These rates of successful case-finding and interview completion are virtually identical to those in the 1996-1997 study.7 The potential respondent was then sent a packet of information containing a letter of introduction, a study explanation sheet, and a postcard to return if they subsequently decided to decline participation. To be included in the sample, the family contact had to indicate that they were involved in the patient's care and decision making in the final month of life. Individuals who were not involved in the patient's care and decision making were excluded from the sample.
Family contacts were interviewed by telephone 2 to 4 months after the death of their family member with the 58-item Barriers Questionnaire, which was developed by the investigators and used in the 1996-1997 survey.7 This time frame was selected to minimize respondent burden and recall error.9 The questionnaire contained the following question regarding the patient's level of pain: “Which best describes the level of comfort of [insert name] in the week before his/her death? Would you say that he/she was: comfortable, in mild pain or distress, in moderate pain or distress, or in severe pain or distress?” Additional questions were asked regarding pain management (table). As in the previous study, graduate research assistants were intensively trained to standardize survey administration, and inter-rater reliability was high (94%).
Study data consisted of information about the decedents from death certificates and from family contacts. Data from death certificates included decedent age, death information (date, place, and cause of death), and an informant's name. The cause of death was classified using the diagnostic categories of the International Classification of Diseases, 9th Revision,10 for the 15 leading causes of death.
In this sample, all the family contacts were related to the decedent by blood or marriage. The relationships to decedents were as follows: wife (n = 37 [36%]), daughter or stepdaughter (n = 20 [19%]), husband (n = 14 [14%]), son or stepson (n = 13 [13%]), daughter-in-law (n = 8 [8%]), and other (for example, sister or mother) (n = 11 [11%]). The top 3 causes of decedents' deaths were cancer (17%), heart disease (16%), and cerebrovascular disease (16%), similar to those in national data. Data about other decedent characteristics and family reports about pain management are presented in the table. Fifty-six family members (54%) reported that their loved one experienced moderate to severe pain in the last week of life. Despite these reports of pain, 92 (89%) described attention to comfort needs as good or excellent.
For in-hospital deaths in early 1997, 33% of family members reported that their loved one had moderate to severe pain in the last week of life. In late 1997, the rate was 57%.7 One year after our first survey, the rate was 54%. Over the 26-month period in which the original and follow-up studies took place, the percentage of in-hospital deaths in Oregon remained stable at approximately 31%.11 The Oregon political climate has been volatile during the past 4 years (figure). Although our study is not robust enough to say that the number of patients dying in pain in hospitals is increasing, we are concerned about this aspect of patient care and the possible effect of political decisions on the levels of pain relief given to patients.
Oregon is the only state in the country whose voters have twice voted to approve an initiative legalizing physician-assisted suicide. The initiative was first passed in 1994 but was put on hold by a court injunction. In the months before the second vote in November 1997, extensive media coverage of the debates about physician-assisted suicide also included related topics such as advance directives, pain management, and hospice care, which educated the public about end-of-life care. The court injunction was quietly lifted in late October 1997, days before voters reaffirmed the initiative a second time.
Following the 1994 affirmative vote on physician-assisted suicide, early data suggested that end-of-life care in Oregon was fairly progressive.12,13,14,15 For example, advance planning for those with serious or life-threatening illness was and still is common.13,16 With the development of comfort-care teams and rising rates of hospice referral,12,13 pain and symptom management also appeared to be improving. In Orgon, the prescription amounts of morphine for all uses was higher in 1998 than it had been in 1997, and Oregon continues to be a leader in the use of morphine per 100,000 population,17 despite family reports of moderate to severe pain for a substantial percentage of dying hospitalized patients.
There may be several reasons that hospitalized patients are still dying in pain. It is possible that the media coverage of late 1997 led families to become more sensitive to the undertreatment of pain and thus to report higher levels of pain when pain was, in reality, no worse. If true, however, the effect did not generalize to families of patients dying in other settings. It may be that hospitals are currently doing more than other settings to heighten family awareness of pain management issues. For example, some Oregon hospitals have posted signs encouraging patients to tell physicians and nurses about their pain, and other hospitals have made pain the fifth vital sign. However, an increase in family expectations about pain management is likely to be only one factor contributing to family reports of increased pain in this subgroup of patients.
Another contributing factor may be that physicians and nurses perceive that they are subject to greater scrutiny from their peers in acute care hospital settings, leaving them more susceptible to pressures that could affect their practice. Other investigators have raised concerns regarding the adverse effects of scrutiny and threats of investigation on physician prescribing of controlled substances.18,19,20 The Institute of Medicine recently reported threats of investigation and legal barriers to prescribing controlled substances as among the reasons physicians undertreat pain.3
Given our difficulty in reaching many family members in our sample, sampling biases may affect the generalizability of our findings. However, unless Oregon death certificates contain informant contact information in the future, it is hard to imagine obtaining a better response rate on a random sample of Oregon death certificates. We did not directly measure patient pain, and several studies suggest that family caregivers have inaccurate perceptions of patient pain.21 We also did not define pain or distress, and respondents may have interpreted these words differently.
Although it is difficult to collect data that will provide proof of connections between complex historical events and patient care (figure), we must not pretend that patient care happens in a vacuum. End-of-life care in Oregon is subject to a variety of complex, volatile political and regulatory forces that are both local and national, and so many families reporting that their loved one experienced moderate to severe pain in the last week of life is worrisome. Since data collection, further changes have taken place. Exploring the effects of environmental factors on pain management may have implications for practice and policy nationwide.
The authors acknowledge the efforts of our research assistants, Zoe Anastas, Diane Baskin, Charlene Bryan, Christie Bernklau Halvor, Bryan Neff, Yon Park, Lisa Fifield-Snodderly, Sandra Tate, and Verna Van-Duynhoven, in conducting the family interviews.
Competing interests: None declared
Funding: This project was supported in part by The Greenwall Foundation, New York, NY; the Meyer Memorial Trust, Portland, OR; and the Open Society Institute Project on Death in America, New York, NY. The views expressed here are those of the authors and do not necessarily represent those of the funding agencies.
See Commentary, p 377.