Volume 40, Issue 6 , Pages 810-820, December 2010
Constipation in Persons Receiving Hospice Care
Article Outline
- Abstract
- Introduction
- Methods
- Results
- Discussion
- Conclusions
- Disclosures and Acknowledgments
- References
- Copyright
Abstract
Context
Symptom burden at the end of life is incompletely understood.
Objectives
To estimate the natural history of constipation and the relation of clinical and demographic characteristics to moderate or severe constipation among persons who received hospice care in the United States starting in 2005.
Methods
Data were obtained from a national provider of hospice pharmacy services and included information about the hospice organization, patient demographics and clinical characteristics, constipation intensity, and drugs prescribed. Hospice nurses assessed patients’ constipation during the previous 24 hours periodically, using a 0–10 numeric rating scale (NRS; 0
=no intensity and 10=worst imaginable; none [NRS 0], mild [NRS 1–3], moderate [NRS 4–6], or severe [NRS 7–10]). Regression models were constructed to identify factors associated with last reported constipation severity scores.
Results
Fifty thousand six hundred forty-one persons received hospice services, had at least two constipation assessments, and had complete clinical and demographic information; 55.3% of these individuals were female, 87.1% were Caucasian, and mean age was 75.9 years. Constipation was assessed a mean of four times per person; 12% of persons had moderate or severe constipation at their first or last assessment, and 19% of persons who reported moderate or severe constipation at the first assessment also had moderate or severe constipation at the last assessment. First constipation and last pain scores, having cancer, and prescription of a laxative were associated with increased likelihood of moderate or severe constipation at the last assessment.
Conclusions
These data provide insight into a common and potentially distressing symptom and also may be useful as process indicators of the quality of hospice care.
Key Words: Constipation, pain, symptom management, hospice, cancer, end of life, opioid
Introduction
Managing symptoms is often challenging for patients and health care providers, particularly at the end of life. Challenges can result from complications associated with life-limiting illnesses and from side effects associated with medications used to manage underlying conditions and associated symptoms. For example, pain is common at the end of life and often suboptimally treated. Although opioid analgesics are the mainstay of pharmacotherapy for moderate-to-severe pain, adverse effects associated with opioids include respiratory depression, sedation, nausea, vomiting, constipation, urinary retention, and itching.1, 2 Tolerance to some of these effects, such as respiratory depression, typically develops quickly, whereas tolerance to constipation generally develops slowly, if at all. If a person becomes constipated, pain may increase, the individuals may become less willing to use analgesics, and in some cases, may need to be hospitalized.
Opioid analgesics cause constipation by acting on the gastrointestinal tract and paraspinal nerve plexi to decrease peristalsis and intestinal secretions, resulting in dry stools.3 Constipation is thought to occur commonly in patients with advanced illness; however, the understanding of the natural history of constipation in persons at the end of life is limited, and few studies have examined this problem. In studies designed to address this gap, Morita et al.4 found that 33% of persons cared for in a Japanese inpatient hospice were constipated on admission and 71% of persons experienced constipation during their hospice experience,4 while Tilden et al.5 and McMillan and Small6 estimated that, among 969 elderly Americans who had died in community settings during the preceding two to five months and were conscious during the last week of life, constipation caused quite a bit or very much distress for 44.4% of participants, and as constipation severity increased, health-related quality of life decreased in persons receiving hospice care.5, 6 Similarly, Fallon and Hanks7 found in a study of 50 persons with advanced cancer that 70% of them were constipated when they were referred to the palliative care team, 23% of these patients were not taking opioids at the time, and after four weeks, 26% were still constipated, and the people who were persistently constipated were statistically significantly less able to function than nonconstipated individuals.
Despite these data, the epidemiology of constipation and its treatment at the end of life remains incompletely understood. Thus, the purpose of this study was to estimate the prevalence and severity of constipation in a large population-based sample of persons who received hospice care. The specific aims of this study were to describe the natural history of constipation in persons who received hospice care, including demographics, primary diagnosis, the severity of constipation on admission to hospice care, numeric rating scale (NRS) intensity scores of constipation severity, laxatives, and opioids prescribed. We also estimated the relationship between constipation and demographic and clinical characteristics of the patients.
Methods
The University of Texas at Austin Investigational Review Board approved the study protocol and waived the requirement for written informed consent. Data for this retrospective cohort study were derived from administrative and clinical information collected by excelleRx, Incorporated (exRx), a national pharmaceutical provider specializing in hospice care under its Hospice Pharmacia division. These data refer to the individual’s most recent episode of hospice care and include information pertaining to the hospice organization, patient demographics and primary diagnoses, medications, and symptom scores, including pain and constipation intensity scores. Data were prospectively collected by pharmacists and technicians during telephone or faxed encounters from hospice nurses and staff. These encounters occurred for a variety of reasons, including hospice admission or discharge, changes in medication orders or site of hospice care, request for a medication refill or renewal, medication consultation, or the patient’s death. Construction of this data set is described in more detail elsewhere.8
The analytic data set included all 50,641 persons in the exRx data set who were admitted to hospice care during the calendar year 2005 and were deceased or discharged alive, had at least two constipation intensity scores, and had complete clinical and demographic data. Study participants were categorized by setting of care; specifically, whether they received hospice care at home, in an inpatient hospice unit, or in a long-term care (LTC, nursing home) facility. Age was calculated relative to the last day of hospice care, to be consistent with the 2000 National Home and Hospice Care Survey (this survey was repeated in 2007, but the results are not yet available).9 Similarly, the length of the hospice care episode was calculated as the number of days from the starting date of care until the date of death or discharge plus one day to ensure that persons who entered and were discharged from hospice or who died on the same day were considered to have received hospice care for one day. The patient’s geographic region was defined using the U.S. Census Bureau map of regions and divisions.10 Primary diagnosis refers to the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) code for the person’s terminal illness.11 Primary diagnosis categories correspond to ICD-9-CM categories, such as infectious disease, digestive system diseases, and so forth. Individuals who had more than one primary diagnosis were categorized as having a primary diagnosis of cancer if any of the diagnostic codes corresponded to a malignant neoplasm. Patients’ constipation during the previous 24 hours was assessed periodically by the hospice nurse using a 0–10 NRS (0=no intensity and 10=worst imaginable). In addition to the numeric rating, pain and constipation intensity ratings on this scale also were categorized as none (NRS 0), mild (NRS 1–3), moderate (NRS 4–6), or severe (NRS 7–10). We assessed opioid analgesic use by counting persons who were prescribed buprenorphine, butorphanol, codeine, fentanyl, hydrocodone, hydromorphone, levorphanol, meperidine, methadone, morphine, nalbuphine, oxycodone, oxymorphone, pentazocine, propoxyphene, or tramadol at least once. Antitussive drugs were excluded from these analyses. Specific details about doses and the order in which opioids were used were not available.
In addition, we constructed a multivariate logistic regression model to help identify factors associated with the conditional probability of being moderately to severely constipated (NRS 4–10) at the last reported assessment, as a function of the person’s first-reported pain and constipation intensity scores; whether the person had cancer; whether the individual was prescribed an opioid, antiemetic, or laxative at least once; and where the person received hospice care. The regression models were adjusted for age, race, sex, and geographic region. All variables were included in the multivariate models at the same time, and all terms were retained in the model, regardless of statistical significance. Because patient care interventions depend in part on the hospice and the team of health care providers at that facility, we used generalized estimating equations to account for clustering by hospice.8 We chose the dependent variable (moderate-to-severe constipation at the last assessment) because last reported intensity scores are likely to be clinically useful. The selection of explanatory variables was based on a theoretical or hypothesized relation to the outcomes being modeled. In particular, we included prescription of antiemetics or laxatives because constipation is associated with nausea, whereas prescription of laxatives may be associated with the probability of having moderate-to-severe constipation at the last report because laxatives are used to prevent and treat constipation. The reference set of covariate values is for a person who was female, Caucasian, younger than 45 years, spent fewer than seven days in hospice, received home care hospice, whose hospice experience ended with the individual’s death, primary disease was cancer, located in the northeastern United States, with no pain or constipation at first assessment, and no opioid, antiemetic, or laxative prescribed. All discharged or deceased persons in the data set with complete clinical and demographic data were included, and analyses were performed using STATA software (StataCorp LP, College Station, TX).
Results
The demographic and clinical characteristics of persons in this sample by their first-reported constipation intensity category are shown in Table 1. In general, study participants were in their mid-70s, female, Caucasian, deceased with a hospice length of service of at least 30 days, and received hospice care at home. Constipation was assessed approximately four times per person, with a median of about 22 days between the first and last assessments, and slightly more than a week between the last assessment and the end of the person’s hospice service. Most patients reported not being constipated at their first assessment. Moderately severe or severe constipation on the first assessment was reported by 12% of study participants and was most prevalent among persons aged 45–64 years, females, Caucasians, persons who received home hospice care, and individuals who were enrolled in hospice at least 30 days.
Table 1. Demographics (n=50,641, %)
| Constipation Reported at First Assessment | ||||
|---|---|---|---|---|
| None (n=39,024, 77.1%) | Mild (n=5,518, 10.9%) | Moderate (n=4,164, 8.2%) | Severe (n=1,935, 3.8%) | |
| Clinical and Demographic Characteristics | n, column % | n, column % | n, column % | n, column % |
| Mean age (median, SD, range), in years | 76.5 (79.3, 14.0, 0.2–104.8) | 74.8 (77.3, 14.1, 0.3–104.5) | 73.6 (76.0, 14.2, 0.2–103.6) | 72.9 (75.3, 14.7, 1.3–103.8) |
| Age group, years | ||||
| <45 (n=1,385) | 1,015 (2.6) | 157 (2.9) | 141 (3.4) | 72 (3.7) |
| 45–64 (n=9,140) | 6,637 (17.0) | 1,133 (20.5) | 903 (21.7) | 467 (24.1) |
| 65–69 (n=4,054) | 3,029 (7.8) | 459 (8.3) | 375 (9.0) | 191 (9.9) |
| 70–74 (n=5,521) | 4,119 (10.6) | 655 (11.9) | 527 (12.7) | 220 (11.4) |
| 75–79 (n=7,358) | 5,619 (14.4) | 812 (14.7) | 655 (15.7) | 272 (14.1) |
| 80–84 (n=8,697) | 6,842 (17.5) | 888 (16.1) | 672 (16.1) | 295 (15.3) |
| ≥85 (n=14,486) | 11,763 (30.1) | 1,414 (25.6) | 891 (21.4) | 418 (21.6) |
| Gender | ||||
| Female (n=28,014) | 21,725 (55.7) | 2,949 (53.4) | 2,241 (53.8) | 1,099 (56.8) |
| Male (n=22,627) | 17,299 (44.3) | 2,569 (46.6) | 1,923 (46.2) | 836 (43.2) |
| Race | ||||
| Asian or Pacific Islander (n=504) | 381 (1.0) | 64 (1.2) | 45 (1.1) | 14 (0.7) |
| Black or African American (n=4,011) | 3,103 (8.0) | 420 (7.6) | 336 (8.1) | 152 (7.9) |
| Caucasian (n=44,104) | 34,008 (87.2) | 4,807 (87.1) | 3,593 (86.3) | 1,696 (87.7) |
| Hispanic (n=1,658) | 1,250 (3.2) | 189 (3.4) | 153 (3.7) | 66 (3.4) |
| Native American (n=66) | 52 (0.1) | 7 (0.1) | 5 (0.1) | 2 (0.1) |
| Other (n=298) | 230 (0.6) | 31 (0.6) | 32 (0.8) | 5 (0.3) |
| Discharge status | ||||
| Deceased (n=43,475) | 33,345 (85.5) | 4,768 (86.4) | 3,652 (87.7) | 1,710 (88.4) |
| Discharged alive (n=7,166) | 5,679 (14.6) | 750 (13.6) | 512 (12.3) | 225 (11.6) |
| Length of hospice service, in days | ||||
| <7 (n=3,003) | 2,368 (6.1) | 264 (4.8) | 233 (5.6) | 138 (7.1) |
| 7–29 (n=16,240) | 12,102 (31.0) | 1,893 (34.3) | 1,507 (36.2) | 738 (38.1) |
| ≥30 (n=31,398) | 24,554 (62.9) | 3,361 (60.9) | 2,424 (58.2) | 1,059 (54.7) |
| Census region | ||||
| Northeast (n=11,247) | 8,741 (22.4) | 1,181 (21.4) | 908 (21.8) | 417 (21.6) |
| Midwest (n=12,071) | 9,531 (24.4) | 1,218 (22.1) | 896 (21.5) | 426 (22.0) |
| South (n=19,062) | 14,519 (37.2) | 2,194 (39.8) | 1,612 (38.7) | 737 (38.1) |
| West (n=8,261) | 6,233 (16.0) | 925 (16.8) | 748 (18.0) | 355 (18.4) |
| Care setting | ||||
| Home care (n=41,540) | 31,125 (79.8) | 4,920 (89.2) | 3,757 (90.2) | 1,738 (89.8) |
| Inpatient (n=374) | 317 (0.8) | 21 (0.4) | 23 (0.6) | 13 (0.7) |
| LTC (n=8,727) | 7,582 (19.4) | 577 (10.5) | 384 (9.2) | 184 (9.5) |
| Last reported constipation | ||||
| None (n=40,570) | 32,347 (82.9) | 4,050 (73.4) | 2,859 (68.7) | 1,314 (67.9) |
| Mild (n=4,433) | 3,037 (7.8) | 646 (11.7) | 527 (12.7) | 223 (11.5) |
| Moderate (n=4,186) | 2,444 (6.3) | 516 (9.4) | 515 (12.4) | 196 (10.1) |
| Severe (n=1,967) | 1,196 (3.1) | 306 (5.6) | 263 (6.3) | 202 (10.4) |
| Opioid prescribed | ||||
| None (n=1,108, 2.2%) | 949 (2.4) | 89 (1.6) | 46 (1.1) | 24 (1.2) |
| Short-acting only (n=26,373) | 21,590 (55.3) | 2,420 (43.9) | 1,662 (39.9) | 701 (36.2) |
| Long acting (n=23,160) | 16,485 (42.2) | 3,009 (54.5) | 2,456 (59.0) | 1210 (62.5) |
| Mean time to first constipation assessment (days, median, SD, range), in days | 4.4 (1.0, 12.7, 1–411) | 3.3 (1.0, 10.8, 1–386) | 3.6 (1.0, 10.9, 1–226) | 3.1 (1.0, 8.8, 1–178) |
| Mean time from first to last constipation assessment (median, SD, range), in days | 44.8 (24.0, 52.6, 1–562) | 41.4 (23.0, 48.0, 1–282) | 38.9 (22.0, 45.9, 1–284) | 37.6 (19.0, 47.3, 1–436) |
| Mean time from last constipation assessment to end of hospice service (median, SD, range), in days | 49.8 (9.0, 95.9, 1–733) | 41.6 (8.0, 86.2, 1–613) | 37.4 (8.0, 81.8, 1–620) | 35.6 (7.0, 77.7, 1–553) |
| Mean no. of times constipation was assessed (median, SD, range) | 3.9 (3.0, 2.7, 2–42) | 4.1 (3.0, 2.8, 2–31) | 4.1 (3.0, 2.7, 2–26) | 4.1 (3.0, 2.9, 2–26) |
Primary diagnoses by first constipation intensity score are presented in Table 2. More than half (53.8%) of study participants had a primary diagnosis of cancer, particularly of the trachea, bronchus, and lung (31.3%) or the digestive organs and peritoneum (26.6%). Moderate-to-severe constipation was most prevalent among persons with cancer, particularly that of the respiratory and intrathoracic organs, the digestive organs and peritoneum, or the genitourinary organs. Among persons with a noncancer primary diagnosis, moderate-to-severe constipation at the first assessment was most common in persons with circulatory system disease, heart disease, or symptoms, signs, and ill-defined conditions.
Table 2. Primary Diagnosis (ICD-9-CM Code) (n=50,641, row %)
| Constipation Reported at First Assessment | ||||
|---|---|---|---|---|
| Primary Diagnosis Category | None (n=39,024, 77.1%) | Mild (n=5,518, 10.9%) | Moderate (n=4,164, 8.2%) | Severe (n=1,935, 3.8%) |
| Infectious and parasitic diseases (001-139, n=335, 0.7%) | 294 (87.8) | 20 (6.0) | 14 (4.2) | 7 (2.1) |
| Malignant neoplasms (140-208, 230-234, n=27,253, 53.8%) | 19,607 (71.9) | 3,519 (12.9) | 2,790 (10.2) | 1,337 (4.9) |
| Digestive organs and peritoneum (150-159, n=7,249, 14.3%) | 5,393 (74.4) | 857 (11.8) | 694 (9.6) | 305 (4.2) |
| Large intestine and rectum (153-154, n=2,724, 5.4%) | 2,104 (77.2) | 300 (11.0) | 215 (7.9) | 105 (3.9) |
| Respiratory and intrathoracic organs (160-165, n=8,803, 17.4%) | 6,238 (70.9) | 1,210 (13.8) | 945 (10.7) | 410 (4.7) |
| Trachea, bronchus, and lung (162, n=8,553, 16.9%) | 6,049 (70.7) | 1,175 (13.7) | 929 (10.9) | 400 (4.7) |
| Bone, connective tissue, skin, and breast (170-176, n=2,837, 5.6%) | 1,959 (69.1) | 387 (13.6) | 328 (11.6) | 163 (5.8) |
| Genitourinary organs (179-189, n=4,392, 8.7%) | 3,043 (69.3) | 578 (13.2) | 493 (11.2) | 278 (6.3) |
| Lymphatic and hematopoietic tissue (196, 200-208, n=1,535, 3.0%) | 1,151 (75.0) | 192 (12.5) | 121 (7.9) | 71 (4.6) |
| All noncancer diagnoses (n=23,388, 46.2%) | 19,417 (83.0) | 1,999 (8.6) | 1,374 (5.9) | 598 (2.6) |
| Endocrine, nutritional, and metabolic disorders and immunity disorders (240-279, n=114, 0.2%) | 94 (82.5) | 9 (7.9) | 3 (2.6) | 8 (7.0) |
| Mental disorders (290-319, n=2,307, 4.6%) | 2,002 (86.8) | 149 (6.5) | 113 (4.9) | 43 (1.9) |
| Diseases of the nervous system and sense organs (320-389, n=3,001, 5.9%) | 2,433 (81.1) | 287 (9.6) | 181 (6.0) | 100 (3.3) |
| Diseases of the circulatory system (390-459, n=7,455, 14.7%) | 6,207 (83.3) | 646 (8.7) | 423 (5.7) | 179 (2.4) |
| Heart disease (391-392, 393-398, 402, 404, 415-416, 420-427, 428-429, n=5,670, 11.2%) | 4,738 (83.6) | 486 (8.6) | 328 (5.8) | 118 (2.1) |
| Diseases of the respiratory system (460-519, n=3,663, 7.2%) | 2,951 (80.6) | 367 (10.0) | 261 (7.1) | 84 (2.3) |
| Chronic obstructive pulmonary disease and allied conditions (490-496, n=3,093, 1.8%) | 2,491 (80.5) | 311 (10.0) | 218 (7.1) | 73 (2.4) |
| Diseases of the digestive system (520-579, n=928, 2.3%) | 811 (87.4) | 47 (5.1) | 51 (5.5) | 19 (2.1) |
| Diseases of the genitourinary system (580-629, n=800, 1.6%) | 683 (85.4) | 57 (7.1) | 44 (5.5) | 16 (2.0) |
| Diseases of the musculoskeletal system and connective tissue (710-739, n=43, 0.1%) | 34 (79.1) | 6 (14.0) | 2 (4.7) | 1 (2.3) |
| Symptoms, signs, and ill-defined conditions (780-799, n=4,254, 8.4%) | 3,516 (82.7) | 362 (8.5) | 250 (5.9) | 126 (3.0) |
Laxatives prescribed are shown in Table 3. Overall, 73.9% of persons had at least one laxative prescribed. Notably, 25.8% of persons who were prescribed at least one opioid were not prescribed a laxative during their hospice care. Some of these patients may have been prescribed opioid analgesics on an as-needed basis only, however, or there may have been other reasons why laxatives were not clinically indicated. Had we been able to assess the data by directions for opioid use, differences may have been apparent. Additionally, among persons prescribed at least one opioid, 4.2% of persons were prescribed only docusate. Oral laxatives were by far the most common type used, and stimulant, mixed content (those containing more than one type of laxative), and softener laxatives were the most commonly prescribed.
Table 3. Laxatives Prescribed (n=50,635, column %)
| Persons Whose Primary Disease Was Not Cancer (n=23,384, 46.2%) | Persons Whose Primary Disease Was Cancer (n=27,251, 53.8%) | ||||
|---|---|---|---|---|---|
| Laxatives Prescribed, by Dosage Form and Type | No Opioid Prescribed (n=950, 4.1%) | Opioid Prescribed (n=22,434, 95.9%) | No Opioid Prescribed (n=152, 0.6%) | Opioid Prescribed (n=27,099, 99.4%) | P-value (Cancer vs. No Cancer) |
| Laxatives by dosage form | |||||
| Any laxative (n=37,412, 73.9%) | 557 (58.6) | 15,839 (70.6) | 79 (52.0) | 20,937 (77.3) | <0.001 |
| Suppositories (n=15,968, 31.5%) | 275 (29.0) | 7,584 (33.8) | 33 (21.7) | 8,076 (29.8) | <0.001 |
| Enemas (n=1,260, 2.5%) | 50 (5.3) | 701 (3.1) | 4 (2.6) | 505 (1.9) | <0.001 |
| Oral dosage forms (n=36,148, 71.4%) | 482 (50.7) | 14,996 (66.8) | 73 (48.0) | 20,597 (76.0) | <0.001 |
| Laxatives, by type | |||||
| Bulk forming (n=961, 1.9%) | 31 (3.3) | 514 (2.3) | 1 (0.7) | 415 (1.5) | <0.001 |
| Castor oil (n=3, <0.1%) | 0 (0.0) | 2 (<0.1) | 0 (0.0) | 1 (<0.1) | 0.599 |
| Lubricant (mineral oil) (n=286, 0.6%) | 9 (1.0) | 130 (0.6) | 2 (1.3) | 145 (0.5) | 0.410 |
| Osmotic (n=9,029, 17.8%) | 101 (10.6) | 3,658 (16.3) | 11 (7.2) | 5,259 (19.4) | <0.001 |
| Polyethylene glycol (n=2,578, 5.1%) | 33 (3.5) | 1,167 (5.2) | 3 (2.0) | 1,375 (5.1) | 0.702 |
| Saline (n=11,852, 23.4%) | 369 (38.8) | 6,040 (26.9) | 37 (24.3) | 5,406 (20.0) | <0.001 |
| Softener (docusate) (n=12,890, 25.5%) | 232 (24.4) | 5,997 (26.7) | 27 (17.8) | 6,634 (24.5) | <0.001 |
| Stimulant (n=32,650, 64.5%) | 431 (45.4) | 13,328 (59.4) | 65 (42.8) | 18,826 (69.5) | <0.001 |
| Mixed content (n=20,891, 41.3%) | 149 (15.7) | 7,139 (31.8) | 34 (22.4) | 13,569 (50.1) | <0.001 |
Estimates of the relation between clinical and demographic characteristics and the likelihood that the person reported having moderate-to-severe constipation at their last assessment are shown in Table 4. In the multivariate analyses, the factors statistically significantly associated with lower likelihood of moderate-to-severe constipation were hospice service of at least 30 days (28% lower), receiving hospice care in either an inpatient (37% lower) or an LTC (45% lower) facility, and the number of times constipation was assessed (2% lower for each additional assessment). Conversely, the factors statistically significantly associated with an increase in the likelihood of moderate or severe last constipation scores were mild, moderate, or severe first constipation scores (likelihood increased by 1.32, 1.64, and 1.83 times, respectively); mild, moderate, or severe last pain scores (likelihood increased by 1.23, 1.37, and 1.68 times, respectively); and whether a laxative was prescribed (likelihood increased by 3.29 times).
Table 4. Conditional Probability of Reporting Being Moderately or Severely Constipated at Last Assessment (n=50,468)
| Characteristic | Adjusted Odds Ratio | 95% Confidence Interval | P-value |
|---|---|---|---|
| First constipation score, by category | |||
| Mild | 1.32 | 1.21, 1.43 | <0.001 |
| Moderate | 1.64 | 1.50, 1.79 | <0.001 |
| Severe | 1.83 | 1.63, 2.06 | <0.001 |
| Length of hospice service, days | |||
| 7–29 | 1.03 | 0.91, 1.17 | 0.629 |
| ≥30 | 0.72 | 0.64, 0.82 | <0.001 |
| Patient type | |||
| Inpatient | 0.63 | 0.42, 0.95 | 0.027 |
| LTC | 0.55 | 0.50, 0.61 | <0.001 |
| Primary disease was cancer | 1.07 | 1.004, 1.15 | 0.037 |
| Opioid prescribed | |||
| Short-acting opioid only | 1.18 | 0.90, 1.56 | 0.238 |
| Long-acting opioid | 1.25 | 0.94, 1.65 | 0.125 |
| First pain category | |||
| Mild | 1.00 | 0.92, 1.08 | 0.921 |
| Moderate | 0.97 | 0.89, 1.05 | 0.455 |
| Severe | 0.95 | 0.87, 1.04 | 0.256 |
| Last pain category | |||
| Mild | 1.23 | 1.14, 1.34 | <0.001 |
| Moderate | 1.37 | 1.27, 1.48 | <0.001 |
| Severe | 1.68 | 1.54, 1.82 | <0.001 |
| Antiemetic prescribed | 1.05 | 0.94, 1.17 | 0.423 |
| Laxative prescribed | 3.29 | 3.00, 3.61 | <0.001 |
| No. of times constipation was assessed | 0.98 | 0.97, 0.99 | 0.004 |
Discussion
The results of this study provide estimates of the natural history of constipation in a large population-based sample of individuals who received hospice care in the United States. To our knowledge, these analyses are the first results of this type. Most persons in this retrospective cohort study were Caucasian females who had a primary diagnosis of cancer and received home-based hospice care. The demographic profile of the individuals in this sample was similar to other published estimates.8 The average length of stay in this study was 93.2 days with a median of 45 days, compared with an average of 47 days with a median of 16 days in the 2000 National Statistics.9 This difference is likely because of the inclusion criteria we used; specifically, individuals who had fewer than two constipation assessments were excluded, thereby removing people whose length of stay was probably short. On average, nurses reported constipation assessments for study participants approximately four times or about every three to four weeks (about every 10 days, based on the median length of stay). This estimate is similar to the frequency of pain assessments that we found in a separate study.8 For approximately half of patients, the last assessment took place within the last week and a half of their hospice experience.
Although published data on the natural history of constipation at the end of life are sparse (a literature search for articles using the terms constipation AND hospice AND epidemiology identified just 10 publications, none of which were specific to constipation), other authors have examined symptom burden cross-sectionally in persons at the end of life. These data tend to differ from our observation that, at the first assessment, 77% of persons reported not being constipated, 23% of people were constipated (at any level of intensity), and 12% of persons reported moderate-to-severe constipation. For example, in 178 persons with cancer who were newly admitted to home hospice care, McMillan and Small6 found that constipation was generally considered mild on admission and was not listed among the most common problems identified by study participants. Notably, however, the authors found that physical symptoms (including constipation) and quality of life, and constipation and total distress were significantly and negatively correlated. Kutner et al.12 also examined symptom burden in hospice patients from providers’ point of view. Among 348 individuals, constipation was reported in 39% of persons and was viewed as severe in 18% of cases when it was present. These authors also found that although constipation was more prevalent in persons with cancer than in individuals with noncancer diagnoses, severe constipation was more prevalent in individuals with noncancer diagnoses. Similarly, Potter et al.13 found that within a sample of 400 individuals (95% of whom had cancer) who had been referred to palliative care services in the United Kingdom (100 to each of four types of services: hospice, community, hospital, and outpatient), 32% overall reported being constipated on the first clinical assessment although prevalence of this symptom was highest in the hospice group (52%) and lowest in the outpatient cohort (17%), suggesting the contribution of overall physical status.6 Last, among 922 hospitalized persons at a comprehensive cancer center, Dhillon et al.14 found that constipation was the presenting problem for 20.3% of persons.
Several factors may have contributed to the differences between other published estimates and the relatively low prevalence of moderate-to-severe constipation we observed. First, the intensity of constipation is likely to be different between persons with different life-limiting illnesses. For example, in the current analysis, 17% of persons with a noncancer primary diagnosis were constipated on the first assessment and 8.5% had moderate-to-severe constipation compared with 28% and 15% of persons with cancer, respectively. Second, the location of care may serve as a proxy for physical condition and symptom severity. In these analyses, 81.8% of patients received home hospice care. Last, the source of the data may contribute to lower prevalence estimates. In this study, hospice nurses reported how the patient would rate their constipation in the past 24 hours. Ideally, the rating was obtained by asking the patient, although it is possible that these data are a combination of patient-, nurse-, and proxy-reported estimates, with underreporting being a possible result of nurse- or proxy-reported symptom intensity. Similarly, if the nurse did not assess the patient’s constipation or the data were otherwise unavailable, it is possible that the score intensity was misclassified as 0 out of 10. We also observed that the severity of constipation improved over time for persons who initially had moderate-to-severe constipation, suggesting that poorly managed symptoms present at the start of the person’s hospice experience were aggressively managed.
In the multivariate logistic regression model we constructed to explore the relation between clinical and demographic characteristics and the likelihood of moderate or severe constipation at the last assessment, the factors associated with a lower likelihood of this outcome were receiving hospice care for at least 30 days, receiving care in a setting other than at home, and having constipation assessed more often. Although laxatives are often prescribed when opioids are initiated, it is also true that unassessed or unreported symptoms cannot be treated. Thus, the latter finding that each assessment was associated with an additional 2% decrease in the likelihood of moderate or severe constipation (i.e., an extra assessment each week of either a mean or median length of stay would be associated with a 26.6% or 12.9% decrease in the likelihood of moderate-to-severe constipation at the last assessment, respectively) suggests the value of providing additional efforts and resources to assess and manage symptoms in an interdisciplinary setting.
In contrast, being constipated on the first assessment, having pain on the last assessment, or having had a laxative prescribed were all statistically significantly associated with an increased likelihood of moderate or severe constipation at the last assessment. Additionally, although prescription of either only short-acting opioids or long-acting opioids was associated with an increased likelihood of the outcome, their effect was not statistically significant. Notably, none of these associations should be interpreted as being causal. In particular, while such a conclusion has face validity, such as the idea that persons who are constipated when they enter hospice are more likely to be moderately to severely constipated at their last assessment, others, such as having been prescribed a laxative, more likely indicate that the drug was being used to treat existing constipation.
Potential limitations of these analyses include confounding, information, and selection biases, which may influence the relation between clinical and demographic data and pain outcomes. We have attempted to account for potential confounding by adjusting for clinical and demographic data in our regression model. Information bias may contribute to distorting our results if there is misclassification or if systematic differences exist in how information was added to the database. Selection bias would be expected to contribute if persons dropped out of the data set differentially between groups. An examination of the frequency of missing race information by hospice in a previous analysis did not indicate particular patterns of missingness.15 Although the clinical and demographic information in this data set reflect what hospice personnel reported, it remains possible that missing information was not collected or was collected but was not available to exRx staff. In addition, some estimates of symptom intensity may have been made by the nurse, rather than reported by the patient. Nurse-reported data may not accurately reflect patients’ actual experience, and proxy reports by nurses could underestimate the degree of constipation experienced by patients, a phenomenon reported in other proxy-reported studies.16, 17 Similarly, as noted above, symptom scores may not be available as the person approaches death. A complete case analysis may also introduce bias if measurement of the covariates and the dropout process are related, as the complete cases cannot then be assumed to be a random sample of the data. Additionally, estimates from the data set used in this study are generally similar to the full exRx data set, 2000 National Statistics, and data from the National Hospice and Palliative Care Organization, whereas the percentage of persons with any type of cancer is higher in the current analyses than in other national data sets of persons who received hospice care.9, 18 As a result, the generalizability of the current results may be limited. Ecologic bias also is possible because we have used population data in this study to try to explain an individual’s experience with constipation during their hospice care. Additionally, although clinical information was limited to the individual’s primary diagnosis, comorbid conditions are well known to contribute to clinical, economic, and patient-reported outcomes and may affect an individual’s symptoms during hospice care. Last, a potential limitation has to do with use of the 0–10 NRS to rate constipation intensity requires that the patient must quantify a complex, subjective state into a single number. Furthermore, it may be difficult to understand how to rate constipation intensity on this type of scale. Categorizing the 0–10 scale into none, mild, moderate, and severe intensity groups may help to address this limitation.
Research to better understand the effect of symptoms in persons at the end of life has many applications, a few of which are suggested by this project. First, as with previous analyses, to address potential biases, more complete and validated data collection is important. Second, these data contribute to quality of care improvement and performance measurement. Examples of this effort are statements from national organizations in the United States, including the Joint Commission, the American Medical Association, and the National Committee for Quality Assurance on performance measures for back pain, cancer, and arthritis, and recent guidelines for the conduct of chronic pain clinical trials.19 Additionally, prevention of opioid-related constipation was an indicator for all three conditions and is currently being tested as a hospice quality indicator as one of 12 advanced illness management measures by IPRO, the quality improvement organization for New York State.20
Conclusions
Suffering at the end of life is common in the United States, and physical symptoms contribute to this burden. In this study, we found that although severe constipation was relatively uncommon, more frequent assessment of constipation was associated with a lower likelihood of experiencing moderate-to-severe constipation. Additionally, although efforts to control pain with opioids may exacerbate constipation, we found that patients’ last-reported constipation scores and last-reported pain scores were related, suggesting that aggressive treatment of each is likely to decrease the other and to improve overall care in persons at the end of life.
Disclosures and Acknowledgments
This study was funded in part by Wyeth, which was acquired by Pfizer Inc. in October 2009. The study was designed with input from each of the authors; however, the funding source had no role in data collection, analysis, interpretation of results, or presentation of information.
References
- . Opioid side effects. clinical updates Pain. 2007;15:1–6
- Management of opioid side-effects in cancer-related and chronic noncancer pain: a systematic review. J Pain. 2003;4:231–256
- Guideline for the management of cancer pain in adults and children. APS Clinical Practice Guidelines Series, No. 3. Glenview, IL: American Pain Society; 2005;
- . Contributing factors to physical symptoms in terminally ill cancer patients. J Pain Symptom Manage. 1999;18:338–346
- . Out-of-hospital death: advance care planning, decedent symptoms, and caregiver burden. J Am Geriatr Soc. 2004;52:532–539
- . Symptom distress and quality of life in patients with cancer newly admitted to hospice home care. Oncol Nurs Forum. 2002;29:1421–1428
- . Morphine, constipation, and performance status in advanced cancer patients. Palliat Med. 1999;13:159–160
- . Pain, demographics, and clinical characteristics in persons who received hospice care in the United States. J Pain Symptom Manage. 2006;32:519–531
- . Characteristics of hospice care discharges and their length of service: United States, 2000. Vital Health Stat. 2003;13:1–36
- . Census regions and divisions of the United States. Available from www.census.govAccessed November 11, 2009
- . Classification of diseases, functioning, and disability. Available from http://www.cdc.gov/nchs/icd.htmAccessed November 11, 2009
- . Symptom burden at the end of life: hospice providers’ perceptions. J Pain Symptom Manage. 2001;21:473–480
- . Symptoms in 400 patients referred to palliative care services: prevalence and patterns. Palliat Med. 2003;17:310–314
- Clinical findings of a palliative care consultation team at a comprehensive cancer center. J Palliat Med. 2008;11:191–197
- . Clinical and demographic characteristics help explain variations in pain at the end of life. J Pain Symptom Manage. 2008;35:10–19
- . Assessing the reliability of patient, nurse, and family caregiver ratings in hospitalized advanced cancer patients. J Clin Oncol. 1999;17:3621–3630
- . Symptom assessment in elderly cancer patients receiving palliative care. Crit Rev Oncol Hematol. 2002;47:281–286
- . Measuring hospice care: the National Hospice and Palliative Care Organization National Hospice data set. J Pain Symptom Manage. 2004;28:316–328
- American Pain Society recommendations for improving the quality of acute and cancer pain management. American Pain Society Quality of Care Task Force. Arch Intern Med. 2005;165:1574–1580
- . Available from http://www.ipro.orgAccessed March 7, 2010
PII: S0885-3924(10)00516-6
doi:10.1016/j.jpainsymman.2010.03.018
© 2010 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved.
Volume 40, Issue 6 , Pages 810-820, December 2010
