Clinical Therapeutics/Volume 36, Number 11, 2014
Original Research
Opioid Interruptions, Pain, and Withdrawal Symptoms in Nursing Home Residents Sarah E. Redding, MD1; Sophia Liu, MD1; William W. Hung, MD, MPH1,2; and Kenneth S. Boockvar, MD, MS1,2,3 1
Icahn School of Medicine at Mount Sinai, New York, New York; 2James J. Peters VA Medical Center, Bronx, New York; and 3Jewish Home Lifecare, New York, New York
ABSTRACT Purpose: Interruptions in opioid use have the potential to cause pain relapse and withdrawal symptoms. The objectives of this study were to observe patterns of opioid interruption during acute illness in nursing home residents and examine associations between interruptions and pain and withdrawal symptoms. Methods: Patients from 3 nursing homes in a metropolitan area who were prescribed opioids were assessed for symptoms of pain and withdrawal by researchers blinded to opioid dosage received, using the Brief Pain Inventory Scale and the Clinical Opioid Withdrawal Scale, respectively, during prespecified time periods. The prespecified time periods were 2 weeks after onset of acute illness (eg, urinary tract infection), and 2 weeks after hospital admission and nursing home readmission, if they occurred. Opioid dosing was recorded and a significant interruption was defined as a complete discontinuation or a reduction in dose of >50% for Z1 day. The covariates age, sex, race, comorbid conditions, initial opioid dose, and initial pain level were recorded. Symptoms pre- and post-opioid interruptions were compared and contrasted with those in a group without opioid interruptions. Findings: Sixty-six patients receiving opioids were followed for a mean of 10.9 months and experienced a total of 104 acute illnesses. During 64 (62%) illnesses, patients experienced any reduction in opioid dosing, with a mean (SD) dose reduction of 63.9%
Accepted for publication October 20, 2014. http://dx.doi.org/10.1016/j.clinthera.2014.10.013 0149-2918/$ - see front matter Published by Elsevier HS Journals, Inc.
November 2014
(29.9%). During 39 (38%) illnesses, patients experienced a significant opioid interruption. In a multivariable model, residence at 1 of the 3 nursing homes was associated with a lower risk of interruption (odds ratio ¼ 0.073; 95% CI, 0.009 to 0.597; P o 0.015). In patients with interruptions, there were statistically insignificant changes in mean (SD) pain score (difference 0.50 [2.66]; 95% CI, 3.16 to 2.16) and withdrawal score (difference 0.91 [3.12]; 95% CI, 4.03 to 2.21) after the interruption as compared with before interruption. However, when compared with patients without interruptions, patients with interruptions experienced larger increases in pain scores during the follow-up periods (difference 0.09 points per day; 95% CI, 0.01 to 0.019; P ¼ 0.08). In particular, patients who received the highest quartile of opioid dose before interruption experienced increases in pain scores over time that were 0.22 points per day larger (95% CI, 0.02 to 0.41; P ¼ 0.03) than those without interruption. Withdrawal scores were not associated with opioid interruption regardless of dose before interruption. Implications: Nursing home patients often experience interruptions in opioid dosing, which can be associated with worse pain, but not withdrawal symptoms, during acute illnesses. Clinicians should be aware of the potential risks and effects of opioid interruptions during acute illnesses in this patient group. (Clin Ther. 2014;36:1555–1563) Published by Elsevier HS Journals, Inc. Scan the QR Code with your phone to obtain FREE ACCESS to the articles featured in the Clinical Therapeutics topical updates or text GS2C65 to 64842. To scan QR Codes your phone must have a QR Code reader installed.
1555
Clinical Therapeutics Key words: opioid, withdrawal, pain management, nursing homes.
INTRODUCTION Among the 1.6 million nursing home patients in the United States, 49%89% suffer from persistent pain, and 38.4% of these receive opioid analgesics.1 During the course of their nursing home stay, many patients experience episodes of acute illness, treated both at the nursing home and in hospital.2 Previous work has shown that medication changes and interruptions are common during these episodes; these interruptions not only represent purposeful changes in response to condition, but also inadvertent changes. For opioid prescriptions, these interruptions can potentially be risky.3 For patients on chronic opioid medications, tolerance and dependence typically occur after 3 weeks of daily use, and withdrawal symptoms can start as soon as 1224 hours after interruptions, sometimes sooner in the case of short-acting opioid medications.4 The severity of withdrawal symptoms depends on the type and frequency of the drug taken, but can initially include a series of flu-like symptoms, including anxiety, restlessness, insomnia, and stomach cramps.5 One small, double-blind, placebo-controlled, crossover study found that the cessation of low-dose chronic opioids in older patients with noncancer pain led to an overall increase in pain, decrease in function, and diminished quality of life, with several patients experiencing symptoms of withdrawal during the abstinence period.6 In the setting of acute illness, withholding opioid medications can lead to relapse in pain or precipitation of withdrawal or other symptoms that can complicate the course of the patients’ illness and recovery.7 Although clinicians might choose to reduce or discontinue chronic opioid prescription for medical reasons, including to minimize polypharmacy, fear of an adverse drug reaction or an acute illness factor, such as altered mental status,5 inadvertent omission can also occur, especially if patients are transferred to another site for care. To our knowledge, little is known about the patterns of opioid interruption in nursing home patients during acute illnesses and the potential consequences of these interruptions. The objectives of this study were to: (1) describe the frequency of opioid interruption in nursing home patients during acute illness, (2) identify clinical and other factors associated with opioid interruption
1556
during acute illness, and (3) examine the association between opioid interruption and symptoms of pain and withdrawal during acute illness. We hypothesized that a combination of patient factors, including demographics, chronic disease burden, and cognitive and physical function, as well as acute clinical factors, can influence opioid interruptions during an acute illness, and that those receiving higher doses of opioid would experience greater effects of interruption if it occurred.
METHODS Design, Setting, and Patients This study was a prospective observational study of nursing home patients designed to describe patterns of medication prescribing for pain, depression, and psychosis in the nursing home, as described previously.8 Patients were enrolled between 2007 and 2009 from 3 nursing homes in the metropolitan New York area, 2 large nonprofit facilities (514 and 816 nursing home beds) and one Department of Veterans Affairs (VA) facility (120 nursing home beds). Patients were included who were prescribed opioids as indicated in the pharmacy record, received at least one opioid dose daily in the 7 days preceding screening according to the medication administration record, and were free from acute illness at the time of screening according to medical and nursing record review. We excluded patients with expected nursing home stays shorter than 2 months, including those admitted for a post acute stay or those on hospice. The Institutional Review Boards at Jewish Home Lifecare and the James J. Peters VA Medical Center in Bronx, NY approved this study. Informed consent was obtained from each patient or a legal surrogate. We collected baseline information on patient physical and cognitive function through interviews with patients, surrogates, and nursing staff. Physical function was assessed using the Activities of Daily Living scale derived from the Minimum Data Set, a federally mandated nursing home resident assessment instrument.9 Cognitive function was assessed and scored using the Cognitive Performance Scale derived from Minimum Data Set.10 We collected the following baseline information through medical record abstraction: demographics; chronic medical conditions; and current medications, including baseline opioid prescription dosage. A count of chronic medical conditions (including coronary disease, congestive heart failure, hypertension, liver disease, peripheral vascular disease,
Volume 36 Number 11
S.E. Redding et al. seizure disorders, dementia, and chronic lung disease) was calculated as a measure of chronic illness burden.2
Surveillance for Acute Illness Surveillance for the occurrence of an acute illness in enrolled patients was performed twice weekly until death or discharge from the nursing home or until study end. The mean period of surveillance among enrolled patients was 10.9 months. Acute illness surveillance was conducted via research staff inquiry of nursing home nursing staff and medical providers using clinical criteria in protocols for physician notification established by the American Medical Directors Association.11 In these protocols, acute medical problems that arose over hours to days, such as chest pain, dyspnea, diarrhea, changes in mental status, signs of hypotension or hypertension, fever, chills, and laboratory abnormalities (such as a drop in hematocrit of >5 points) were considered indicators of an acute change in condition. When an acute illness occurred, we collected from the medical record the diagnosis of the acute illness, whether it resulted in hospital transfer, and its severity, using the Inpatient Physiologic Failure Score, which scores illness severity using vital signs and laboratory tests.12
Opioid Prescribing We recorded daily opioid prescribing information during each of the follow-up periods: 14 days after acute illness onset; if hospitalized, during the hospitalization; and if readmitted to the nursing home, 14 days after nursing home readmission. Opioid medication dosages were converted into morphine milligram equivalents using published dose equivalency tables.16–18 A significant opioid interruption was defined as 1 or more continuous days of complete discontinuation in dosing or a dose reduction >50%. When an interruption was identified, the mean dose in the 7 days preceding the interruption start date was calculated. The percent dose decrease for each medication interruption then was calculated. We conducted a medical record review to ascertain the cause of the opioid dosing interruption among patients who experienced an interruption at one of the study nursing homes. The following information was extracted from the record: whether a clinical reason was documented by care providers, whether relapse or withdrawal symptoms were observed by staff providers and documented in the chart, and whether the interruption could have plausibly caused the relapse or withdrawal symptoms if they occurred, using a scale adapted from Naranjo et al. 19
Follow-Up and Symptom Measurement We followed patients for 14 days after illness onset. The majority of illnesses were managed solely at the nursing home. Patients who were admitted to 2 associated study hospitals were followed during the hospital stay and, if readmitted to the nursing home, followed for 14 days after nursing home readmission. During these follow-up periods, patients were assessed every other day for signs of pain using the Brief Pain Inventory (010, 10 being the worst pain)13 and signs of autonomic sympathetic activation (eg, tachycardia, sweating, restlessness, mydriasis) using the Clinical Opiate Withdrawal Scale.14 In some cases, patients were unable to complete the Brief Pain Inventory and pain was assessed in these patients using the McGill Present Pain Intensity Scale, which has previously been shown to have a high completion rate for communicatively impaired patients.15 Patient assessments were done by a trained research assistant without knowledge of opioid medications received by the patient during the illness (ie, blinded to whether opioid medication receipt was interrupted or not).
November 2014
Data Analysis The unit of analysis was episode of acute illness, and more than one acute illness was allowed per patient. Descriptive statistics of characteristics of patients, illnesses, and opioid prescribing patterns are presented. A multivariable logistic regression model was estimated with characteristics of patients and illnesses as independent variables and opioid interruption (yes or no) as dependent variable. To examine potential effects of opioid interruption, we calculated changes in pain and withdrawal scores within patients during interruption as compared with before interruption. Then, these changes were compared in 2 groups of patients using the standard t test: those who had an interruption in their opioid dosage during the course of their acute illness versus those that did not. Because, on average, opioid interruption occurred on day 5 of acute illness, in the noninterruption group we calculated changes in symptoms before and after this day. Pain and withdrawal scores for both groups were also graphed against time, with
1557
Clinical Therapeutics the day of interruption (or day 5 of acute illness in the noninterruption group) aligned as time 0. Finally, we used a multivariable mixed linear model with random intercept and slope to examine the independent effect of opioid interruption on pain and withdrawal scores, accounting for repeated measures and using a time-by-interruption interaction term to compare those who had an interruption with those who did not. We adjusted for the effects of baseline scores and hospitalization in our models. We further explored if the effect of drug interruption was different among those with higher doses of opioids at baseline by stratifying for highest quartile of opioid (Z49.3
mean daily morphine equivalents). All analyses were performed using IBM SPSS software, version 21 (IBM SPSS, Armonk, New York), SAS software, version 9.3 (SAS Institute, Cary, North Carolina), and Stata software, version 12 (Stata Corp, College Station, Texas).
RESULTS Sixty-six nursing home patients enrolled in the study. Mean age was 74 years, 61% were male, and 36% were functionally independent or needed only supervision for their activities of daily living. Mean duration of nursing home stay before enrollment was 2.1 years. Enrolled patients received a mean of 8 non-
Table I. Characteristics of patients.
Characteristic
Nursing Home
Nursing Home Patients
Patients Prescribed Opioids
Prescribed Opioids Who Experienced Acute Illness
Patients, n Age at enrollment, y, mean (SD) [range] Sex, male, n (%)
66
45
73.9 (11.9) [4696] 40 (60.6)
75.1 (12.2) [5395] 25 (55.6)
Race, n (%) White Black or African American
39 (59.1) 22 (33.3)
30 (66.7) 13 (28.9)
Unknown or not reported
4 (6.1)
2 (4.4)
58 (87.9)
42 (93.3)
Ethnicity, n (%) Not Hispanic or Latino Hispanic or Latino Unknown Cognitive function (MDS-CPS) Intact or Borderline Intact (01) Mild or moderate impairment (23) Moderate severe to severe (45)
7 (10.6)
2 (4.4)
1 (1.5)
1 (2.2)
51 (77.3)
35 (77.8)
7 (10.6) 5 (7.5)
6 (13.3) 3 (6.6)
24 (36.7) 15 (22.7)
15 (33.3) 11 (24.3)
Physical impairment (ADL) Independent or supervision (01) Limited or extensive (23) Dependent or total (46) Duration of NH stay, d, mean (range) No. of nonopioid medications, mean (SD)
24 (36.3) 759.3 (134895) 7.76 (3.12)
19 (42.2) 752.09 (204895) 7.82 (3.03)
No. of chronic medical conditions, mean (SD)
3.71 (1.65)
3.91 (1.68)
Hospital days in previous year, mean (SD) No. of illnesses
19.0 (34.8)
19.16 (37.80) 104
Illness severity score (IPFS), mean (SD) [range 039; higher ¼ worse]
2.68 (4.06) [015]
Hospital admission, n (%) If admitted, days of hospital stay, mean (SD) [range]
24 (23) 8.38 (7.85) [135]
ADL ¼ activities of daily living; CPS ¼ Cognitive Performance Scale; IPFS ¼ Inpatient Physiologic Failure Score; MDS ¼ Minimum Data Set; NH ¼ nursing home.
1558
Volume 36 Number 11
S.E. Redding et al. opioid medications, and had a mean of 3.7 chronic conditions. Of the 66 patients receiving opioids that were followed for a mean of 10.9 months, 45 experienced a total of 104 acute illnesses. The majority (77%) of these acute illnesses were treated in the nursing home. The most common diagnoses were cellulitis, urinary tract infection, upper and lower respiratory infections, chronic obstructive pulmonary disease exacerbation, sepsis, and dehydration. Characteristics of the subgroup that became acutely ill during the study period were similar to those of the group as a whole (Table I). During 64 illnesses (62%), patients experienced any reduction in their opioid medication dosage. The mean dose reduction was 63.9%, from 63 daily morphine milligram equivalents to 29. During 39 illnesses (38%), patients experienced a significant opioid interruption (complete discontinuation or >50% dose reduction). Significant interruptions occurred on average on day 5 of acute illness (mean [SD] days, 5.06 [4.02]) and lasted for a mean of 3 days (range 118 days) (Table II). The most frequently interrupted medication was oxycodone (50% of all interruptions). Interruptions occurred most frequently in the nursing home (63%), and the
Table II. Characteristics of opioid interruptions. Episodes of illness among patients 104 prescribed opioids, n Episodes with opioid interruption, n (%) 64 (62) Days from illness onset to interruption, mean (SD) 5.06 (4.02) Duration of interruption, mean days (range) 3 (118) Complete discontinuation, n (%) 22 (21) Daily morphine milligram equivalents, mean (SD) Before interruption 62.97 (87.50) During interruption 29.12 (48.55) % Dose reduction, mean (SD) 63.9 (29.9) Interruption location, n (%) In nursing home (without or before 40 (63) hospital transfer) In hospital 14 (22) In nursing home after hospital discharge 9 (14) Interrupted medication, n Oxycodone 32 Methadone 5 Hydrocodone 3 Codeine 1 Hydromorphone 1 Multiple opioids 20
November 2014
remainder occurred in hospital (22%) or after readmission to the nursing home from the hospital (14%). In bivariate analysis, increasing age was found to be associated with a lower risk of opioid interruption (odds ratio [OR] ¼ 0.947; 95% CI, 0.8980.999; P ¼ 0.04). In a multivariable analysis, only residence at one nursing home (the VA site) was associated with a lower risk of interruption (OR ¼ 0.073; 95% CI, 0.0090.597; P o 0.015) (Table III). The mean baseline pain score was 5.9. During illnesses in which patients experienced a significant interruption in opioid dosing, there were insignificant changes in mean (SD) pain score (difference 0.50 [2.66]; 95% CI, 3.16 to 2.16) and withdrawal score (difference 0.91 [3.12]; 95% Cl, 4.03 to 2.21) within those patients as compared with preinterruption. Similarly, during illnesses in which patients experienced no significant reduction in opioid dosing, there were no significant changes in mean (SD) pain score (difference 0.96 [2.69]; 95% CI, 3.65 to 1.74) or withdrawal score (difference 0.34 [2.16]; 95% CI, 1.82 to 2.5) within those patients before and after day 5 of acute illness (Figures 1 and 2). However, when compared with patients without interruptions, patients with interruptions had larger increases in pain scores over time (difference 0.09 points per day; 95% CI, 0.01 to 0.019; P = 0.08). In particular, among patients who received the highest quartile of opioid dose, those with interruptions experienced increases in pain scores over time that were 0.22 points per day larger (95% CI, 0.02 to 0.41; P=0.03) than those without interruption. Withdrawal scores were not associated with drug interruption over time (point estimate 0.12; 95% CI, 0.07 to 0.30; P = 0.21), regardless of opioid dose before interruption. Of 18 charts reviewed to ascertain the reasons for opioid interruption, 7 (39%) had clear documentation and a clinical reason for opioid interruption, the most common being altered mental status. On the other hand, 11 (61%) had no documentation for the reason for interruption. In many of these cases, the interruption occurred during a transition of care such as change in staffing, transfer to hospital, or transfer to a different unit within the nursing home.
DISCUSSION In this study, we found that a large proportion of nursing home residents experienced reductions in
1559
Clinical Therapeutics
Table III. Factors associated with opioid interruption. Independent Variable Patient characteristics Age Chronic medical conditions Physical function Cognitive function No. medications Opioid dose (morphine milligram equivalents) Nursing home site Illness characteristics Illness severity Hospitalization (yes or no) Increased depression –– Increased pain Change in ability to communicate *
Bivariate P Value
Multivariate P Value
0.040* 0.911 0.559 0.821 0.407 0.133
0.304 0.654 0.826 0.694 0.127 0.110
0.963 0.896 0.947 0.889 1.346 1.010
0.440
0.015*
0.073 (0.0090.597)
0.274 0.180 0.898 –– 0.913 0.547
Multivariate Odds of Interruption (95% CI)
0.188 0.116 0.999 –– 0.683 0.309
(0.8971.034) (0.5531.450) (0.5841.537) (0.4951.597) (0.9191.970) (0.9981.023)
0.813 (0.5981.106) 0.082 (0.0041.854) –– –– (15.242) 0.719 (0.1473.509) 0.385 (0.0612.422)
P r .05
12 10 8 Mean BPI Scores
6 4 2 0
–10
–5
5
10
15
20
Time of Interruption –2 –4
Time (days)
Figure 1. Mean pain scores (Brief Pain Inventory [BPI]) versus time (days) after interruption. Squares and dashed lines ¼ group with significant opioid interruption. Triangles with filled line ¼ group with no significant opioid interruption.
1560
Volume 36 Number 11
S.E. Redding et al.
12
10
Mean COWS Score
8
6
4
2
–5
0 5 Time of Interruption
10
15
20
–2 Time (days)
Figure 2. Mean withdrawal scores (Clinical Opiate Withdrawal Scale [COWS]) versus time (days) after interruption. Squares and dashed lines ¼ group with significant opioid interruption. Triangles with filled line ¼ group with no significant opioid interruption.
opioid prescription dosage during acute illness (62% of illnesses), and many of these events were complete discontinuations or large (>50%) dose reductions. These interruptions occurred regardless of whether the patient was transferred to a hospital. We found that during these interruptions, there were small, insignificant changes in pain scores or withdrawal symptoms within patients when compared with patients’ preinterruption scores. However, when compared with patients who did not have interruptions, those with significant interruptions had larger pain increases over time. The difference in pain trajectories was particularly significant among those receiving higher preinterruption doses of opioid medications. Among these patients, those with opioid interruption had pain that increased by 0.22 points per day more than those without interruption, which, during the course of a mean follow-up of 7 days, could result in pain that is 1.5 points (0.22 7) higher on a 10-point scale. This could represent a difference between experience of mild and moderate pain, or between moderate and severe pain, a clinically significant difference for patients. In addition, for US nursing facilities, this could impact the Centers for Medicare
November 2014
and Medicaid Services publically reported quality measure “percentage of residents with moderate-tosevere pain.” Our finding that there is a high rate of opioid interruption is consistent with earlier work that has examined medication interruption patterns during transfer from nursing homes to the hospitals,20 although rather surprisingly, whether a resident was transferred to the hospital or not was not associated with interruptions. This suggests that, in many cases, interruptions are driven by clinical factors and changes in management within the nursing home itself. Our finding that patients with large interruptions had higher pain scores than those without interruption is consistent with our hypothesis that interruptions in opioid medications can lead to worse pain control, an association that was even more prominent among those receiving higher doses of opioid medications in our study. We did not find that patients had worse withdrawal symptoms with large interruptions, which could be due to the difficulty of detecting changes in autonomic sympathetic activity in a study population that had high rates of chronic disease, medication use, and disability.
1561
Clinical Therapeutics We also found that among factors tested for their association with interruptions, only the nursing home site (the VA site) was associated with less interruption. Of note, patients at the VA nursing home, if needed, were admitted to the VA hospital in the same facility, which shared a similar electronic medical record and was served by the same pharmacy. There might be factors associated with the VA site of care, such as a shared electronic medical record, that may be associated with a lower risk of interruption. This is important because our review for reasons for interruptions found that the majority of cases did not have a documented clinical reason, and may have been precipitated by handoffs in care. Our finding is consistent with other studies on medication omissions in hospitals21 that highlight that clinicians need to improve their documenting of reasons for interruptions and to minimize unnecessary interruptions. There were several limitations to this study. Although our study had a structured protocol for monitoring for acute illnesses, it is possible that short 1- or 2-day delays in identifying acute illnesses could have occurred. Additionally, outcome assessments were performed every other day; therefore, it might have been possible to miss pain or withdrawal symptoms that occurred between assessments. We anticipate that the effects of these off days would be minimal, given that most interruptions occurred at a mean of 5 days after onset of the illness. Also, our power to observe smaller associations was limited by the small size of the cohort. The small sample size was in part a result of a relatively low prevalence of patients receiving opioid analgesics in the study setting, challenges with obtaining informed consent in the nursing home, and a lower than expected incidence of acute illness. Finally, we have missing data on a small proportion of patients (3%) that had hospitalizations at nonstudy hospitals.
CONCLUSIONS Interruptions in opioid medications occur frequently among nursing home patients during acute illness episodes. Clinicians should be wary of the possibility of worsened pain during acute illness if opioid medications are interrupted. Careful medication reconciliation is essential and clinicians need to consider the reasons for and the risks of opioid interruption in this setting.
1562
ACKNOWLEDGMENTS The authors are grateful to Daniel Signor for his role as database and statistical programmer and to Jessica Singleton, Jennifer Kwak, and Julia Siegel Breton, MD for their role in data collection. Dr. Redding contributed to data collection, data analysis, data interpretation, and manuscript writing; Dr. Liu contributed to study design, data collection and data analysis; Dr. Hung contributed to data analysis, data interpretation, and manuscript writing; Dr. Boockvar contributed to study design, data collection, analysis and interpretation, and manuscript writing.
CONFLICTS OF INTEREST This study was supported by Veterans Affairs Health Services Research and Development Service grants RCD 03-027-1 and REA 08-260. The study sponsor had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. One author (K. Boockvar) is also supported by the Greenwall Foundation. The contents do not represent the views of the US Department of Veterans Affairs or the United States Government.
REFERENCES 1. Won AB, Lapane KL, Vallow S, et al. Persistent nonmalignant pain and analgesic prescribing patterns in elderly nursing home residents. J Am Geriatr Soc. 2004;52:867–874. 2. Boockvar KS, Gruber-Baldini AL, Burton L, et al. Outcomes of infection in nursing home residents with and without early hospital transfer. J Am Geriatr Soc. 2005;53:590–596. 3. Suchanek M, Boockvar KS, Morrison RS, Fried T. Discontinuation of standing opioid orders in individuals transferred form nursing home to hospital. J Am Geriatr Soc. 2006;54:S204–S229. 4. Jaffe JH. Misinformation: euphoria and addiction. In: Hill CS Jr, Fields WS, eds. Advances in Pain Research and Therapy, Vol 11. New York: Raven Press; 1989:163–173. 5. Kosten TR, O’Conner PG. Management of Drug and Alcohol Withdrawal. New Engl J Med. 2003;348:1786– 1795. 6. Cowan DT, Wilson-Barnett J, Griffiths P, et al. A randomized, double-blind, placebo-controlled, cross-over pilot study to assess the effects of long-term opioid drug consumption and subsequent abstinence in chronic
Volume 36 Number 11
S.E. Redding et al.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
noncancer pain patients receiving controlled-release morphine. Pain Med. 2005;6:113–121. Gerety MB, Cornell JE, Plichta DT, Eimer M. Adverse Events Related to Drugs and Drug Withdrawal in Nursing Home Residents. J Am Geriatr Soc. 1993;41:1326–1332. Hung WW, Liu S, Boockvar KS. A prospective study of symptoms, function, and medication use during acute illness in nursing home residents: Design, rationale and cohort description. BMC Geriatr. 2010;10:47. Morris JN, Fries BE, Morris SA. Scaling ADLs within the MDS. J Gerontol A Biol Sci Med Sci. 1999;54: M546–M553. Morris JN, Fries BE, Mehr DR, et al. MDS Cognitive Performance Scale. J Gerontol. 1994;49:M174– M182. Protocols for Physician Notification. In: Levenson S, Vance J, eds. Assessing Patients and Collecting Data on Nursing Facility Patients. Columbia, MD: American Medical Directors Association; 2004. Gray LK, Smyth KA, Palmer RM, et al. Heterogeneity in Older People: Examining Physiologic Failure, Age, and Comorbidity. J Am Geriatr Soc. 2002;50:1955–1961. Mendoza T, Mayne T, Rublee D, Cleeland C. Reliability and validity of a modified Brief Pain Inventory short form in patients with osteoarthritis. Eur J Pain. 2006;10: 353, e361. Wesson DR, Ling W. The Clinical Opiate Withdrawal Scale (COWS). J Psychoactive Drugs. 2003;35:253–259. Pautex S, Herrmann F, Le Lous P, et al. Feasibility and Reliability of Four Pain Self-Assessment Scales and Correlation with an Observational Rating Scale in Hospitalized Elderly Demented Patients. J Gerontol A Biol Sci Med Sci. 2005;60: 524–529. AGS Panel on Chronic Pain in Older Persons. The management
November 2014
17.
18.
19.
20.
21.
of chronic pain in older persons. American Geriatrics Society. J Am Geriatr Soc. 1988;46:635–651. Patanwala AE, Duby J, Waters D, Erstad BL. Opioid conversions in acute care. Ann Pharmacother. 2007; 41:255–266. Ripamonti C, Groff L, Brunelli C, et al. Switching from morphine to oral methadone in treating cancer pain: what is the equianalgesic dose ratio? J Clin Oncol. 1998; 15:3216–3221. Naranjo CA, Busto U, Sellers EM, et al. A method for estimating the probability of adverse reactions. Clin Pharmacol Ther. 1981;30:239– 245. Boockvar K, Fishman E, Kyriacou CK, et al. Adverse events due to discontinuations in drug use and dose changes in patients transferred between acute and longterm care facilities. Arch Intern Med. 2004;164:545–550. Warne S, Endacott R, Ryan H, et al. Non-therapeutic omission of medications in acutely ill patients. Nurs Crit Care. 2010;15:112– 117.
Address correspondence to: Kenneth Boockvar, MD, James J Peters VA Medical Center, 130 West Kingsbridge Road, Bronx, New York 10468. E-mail: [email protected]
1563
Original Research
Opioid Interruptions, Pain, and Withdrawal Symptoms in Nursing Home Residents Sarah E. Redding, MD1; Sophia Liu, MD1; William W. Hung, MD, MPH1,2; and Kenneth S. Boockvar, MD, MS1,2,3 1
Icahn School of Medicine at Mount Sinai, New York, New York; 2James J. Peters VA Medical Center, Bronx, New York; and 3Jewish Home Lifecare, New York, New York
ABSTRACT Purpose: Interruptions in opioid use have the potential to cause pain relapse and withdrawal symptoms. The objectives of this study were to observe patterns of opioid interruption during acute illness in nursing home residents and examine associations between interruptions and pain and withdrawal symptoms. Methods: Patients from 3 nursing homes in a metropolitan area who were prescribed opioids were assessed for symptoms of pain and withdrawal by researchers blinded to opioid dosage received, using the Brief Pain Inventory Scale and the Clinical Opioid Withdrawal Scale, respectively, during prespecified time periods. The prespecified time periods were 2 weeks after onset of acute illness (eg, urinary tract infection), and 2 weeks after hospital admission and nursing home readmission, if they occurred. Opioid dosing was recorded and a significant interruption was defined as a complete discontinuation or a reduction in dose of >50% for Z1 day. The covariates age, sex, race, comorbid conditions, initial opioid dose, and initial pain level were recorded. Symptoms pre- and post-opioid interruptions were compared and contrasted with those in a group without opioid interruptions. Findings: Sixty-six patients receiving opioids were followed for a mean of 10.9 months and experienced a total of 104 acute illnesses. During 64 (62%) illnesses, patients experienced any reduction in opioid dosing, with a mean (SD) dose reduction of 63.9%
Accepted for publication October 20, 2014. http://dx.doi.org/10.1016/j.clinthera.2014.10.013 0149-2918/$ - see front matter Published by Elsevier HS Journals, Inc.
November 2014
(29.9%). During 39 (38%) illnesses, patients experienced a significant opioid interruption. In a multivariable model, residence at 1 of the 3 nursing homes was associated with a lower risk of interruption (odds ratio ¼ 0.073; 95% CI, 0.009 to 0.597; P o 0.015). In patients with interruptions, there were statistically insignificant changes in mean (SD) pain score (difference 0.50 [2.66]; 95% CI, 3.16 to 2.16) and withdrawal score (difference 0.91 [3.12]; 95% CI, 4.03 to 2.21) after the interruption as compared with before interruption. However, when compared with patients without interruptions, patients with interruptions experienced larger increases in pain scores during the follow-up periods (difference 0.09 points per day; 95% CI, 0.01 to 0.019; P ¼ 0.08). In particular, patients who received the highest quartile of opioid dose before interruption experienced increases in pain scores over time that were 0.22 points per day larger (95% CI, 0.02 to 0.41; P ¼ 0.03) than those without interruption. Withdrawal scores were not associated with opioid interruption regardless of dose before interruption. Implications: Nursing home patients often experience interruptions in opioid dosing, which can be associated with worse pain, but not withdrawal symptoms, during acute illnesses. Clinicians should be aware of the potential risks and effects of opioid interruptions during acute illnesses in this patient group. (Clin Ther. 2014;36:1555–1563) Published by Elsevier HS Journals, Inc. Scan the QR Code with your phone to obtain FREE ACCESS to the articles featured in the Clinical Therapeutics topical updates or text GS2C65 to 64842. To scan QR Codes your phone must have a QR Code reader installed.
1555
Clinical Therapeutics Key words: opioid, withdrawal, pain management, nursing homes.
INTRODUCTION Among the 1.6 million nursing home patients in the United States, 49%89% suffer from persistent pain, and 38.4% of these receive opioid analgesics.1 During the course of their nursing home stay, many patients experience episodes of acute illness, treated both at the nursing home and in hospital.2 Previous work has shown that medication changes and interruptions are common during these episodes; these interruptions not only represent purposeful changes in response to condition, but also inadvertent changes. For opioid prescriptions, these interruptions can potentially be risky.3 For patients on chronic opioid medications, tolerance and dependence typically occur after 3 weeks of daily use, and withdrawal symptoms can start as soon as 1224 hours after interruptions, sometimes sooner in the case of short-acting opioid medications.4 The severity of withdrawal symptoms depends on the type and frequency of the drug taken, but can initially include a series of flu-like symptoms, including anxiety, restlessness, insomnia, and stomach cramps.5 One small, double-blind, placebo-controlled, crossover study found that the cessation of low-dose chronic opioids in older patients with noncancer pain led to an overall increase in pain, decrease in function, and diminished quality of life, with several patients experiencing symptoms of withdrawal during the abstinence period.6 In the setting of acute illness, withholding opioid medications can lead to relapse in pain or precipitation of withdrawal or other symptoms that can complicate the course of the patients’ illness and recovery.7 Although clinicians might choose to reduce or discontinue chronic opioid prescription for medical reasons, including to minimize polypharmacy, fear of an adverse drug reaction or an acute illness factor, such as altered mental status,5 inadvertent omission can also occur, especially if patients are transferred to another site for care. To our knowledge, little is known about the patterns of opioid interruption in nursing home patients during acute illnesses and the potential consequences of these interruptions. The objectives of this study were to: (1) describe the frequency of opioid interruption in nursing home patients during acute illness, (2) identify clinical and other factors associated with opioid interruption
1556
during acute illness, and (3) examine the association between opioid interruption and symptoms of pain and withdrawal during acute illness. We hypothesized that a combination of patient factors, including demographics, chronic disease burden, and cognitive and physical function, as well as acute clinical factors, can influence opioid interruptions during an acute illness, and that those receiving higher doses of opioid would experience greater effects of interruption if it occurred.
METHODS Design, Setting, and Patients This study was a prospective observational study of nursing home patients designed to describe patterns of medication prescribing for pain, depression, and psychosis in the nursing home, as described previously.8 Patients were enrolled between 2007 and 2009 from 3 nursing homes in the metropolitan New York area, 2 large nonprofit facilities (514 and 816 nursing home beds) and one Department of Veterans Affairs (VA) facility (120 nursing home beds). Patients were included who were prescribed opioids as indicated in the pharmacy record, received at least one opioid dose daily in the 7 days preceding screening according to the medication administration record, and were free from acute illness at the time of screening according to medical and nursing record review. We excluded patients with expected nursing home stays shorter than 2 months, including those admitted for a post acute stay or those on hospice. The Institutional Review Boards at Jewish Home Lifecare and the James J. Peters VA Medical Center in Bronx, NY approved this study. Informed consent was obtained from each patient or a legal surrogate. We collected baseline information on patient physical and cognitive function through interviews with patients, surrogates, and nursing staff. Physical function was assessed using the Activities of Daily Living scale derived from the Minimum Data Set, a federally mandated nursing home resident assessment instrument.9 Cognitive function was assessed and scored using the Cognitive Performance Scale derived from Minimum Data Set.10 We collected the following baseline information through medical record abstraction: demographics; chronic medical conditions; and current medications, including baseline opioid prescription dosage. A count of chronic medical conditions (including coronary disease, congestive heart failure, hypertension, liver disease, peripheral vascular disease,
Volume 36 Number 11
S.E. Redding et al. seizure disorders, dementia, and chronic lung disease) was calculated as a measure of chronic illness burden.2
Surveillance for Acute Illness Surveillance for the occurrence of an acute illness in enrolled patients was performed twice weekly until death or discharge from the nursing home or until study end. The mean period of surveillance among enrolled patients was 10.9 months. Acute illness surveillance was conducted via research staff inquiry of nursing home nursing staff and medical providers using clinical criteria in protocols for physician notification established by the American Medical Directors Association.11 In these protocols, acute medical problems that arose over hours to days, such as chest pain, dyspnea, diarrhea, changes in mental status, signs of hypotension or hypertension, fever, chills, and laboratory abnormalities (such as a drop in hematocrit of >5 points) were considered indicators of an acute change in condition. When an acute illness occurred, we collected from the medical record the diagnosis of the acute illness, whether it resulted in hospital transfer, and its severity, using the Inpatient Physiologic Failure Score, which scores illness severity using vital signs and laboratory tests.12
Opioid Prescribing We recorded daily opioid prescribing information during each of the follow-up periods: 14 days after acute illness onset; if hospitalized, during the hospitalization; and if readmitted to the nursing home, 14 days after nursing home readmission. Opioid medication dosages were converted into morphine milligram equivalents using published dose equivalency tables.16–18 A significant opioid interruption was defined as 1 or more continuous days of complete discontinuation in dosing or a dose reduction >50%. When an interruption was identified, the mean dose in the 7 days preceding the interruption start date was calculated. The percent dose decrease for each medication interruption then was calculated. We conducted a medical record review to ascertain the cause of the opioid dosing interruption among patients who experienced an interruption at one of the study nursing homes. The following information was extracted from the record: whether a clinical reason was documented by care providers, whether relapse or withdrawal symptoms were observed by staff providers and documented in the chart, and whether the interruption could have plausibly caused the relapse or withdrawal symptoms if they occurred, using a scale adapted from Naranjo et al. 19
Follow-Up and Symptom Measurement We followed patients for 14 days after illness onset. The majority of illnesses were managed solely at the nursing home. Patients who were admitted to 2 associated study hospitals were followed during the hospital stay and, if readmitted to the nursing home, followed for 14 days after nursing home readmission. During these follow-up periods, patients were assessed every other day for signs of pain using the Brief Pain Inventory (010, 10 being the worst pain)13 and signs of autonomic sympathetic activation (eg, tachycardia, sweating, restlessness, mydriasis) using the Clinical Opiate Withdrawal Scale.14 In some cases, patients were unable to complete the Brief Pain Inventory and pain was assessed in these patients using the McGill Present Pain Intensity Scale, which has previously been shown to have a high completion rate for communicatively impaired patients.15 Patient assessments were done by a trained research assistant without knowledge of opioid medications received by the patient during the illness (ie, blinded to whether opioid medication receipt was interrupted or not).
November 2014
Data Analysis The unit of analysis was episode of acute illness, and more than one acute illness was allowed per patient. Descriptive statistics of characteristics of patients, illnesses, and opioid prescribing patterns are presented. A multivariable logistic regression model was estimated with characteristics of patients and illnesses as independent variables and opioid interruption (yes or no) as dependent variable. To examine potential effects of opioid interruption, we calculated changes in pain and withdrawal scores within patients during interruption as compared with before interruption. Then, these changes were compared in 2 groups of patients using the standard t test: those who had an interruption in their opioid dosage during the course of their acute illness versus those that did not. Because, on average, opioid interruption occurred on day 5 of acute illness, in the noninterruption group we calculated changes in symptoms before and after this day. Pain and withdrawal scores for both groups were also graphed against time, with
1557
Clinical Therapeutics the day of interruption (or day 5 of acute illness in the noninterruption group) aligned as time 0. Finally, we used a multivariable mixed linear model with random intercept and slope to examine the independent effect of opioid interruption on pain and withdrawal scores, accounting for repeated measures and using a time-by-interruption interaction term to compare those who had an interruption with those who did not. We adjusted for the effects of baseline scores and hospitalization in our models. We further explored if the effect of drug interruption was different among those with higher doses of opioids at baseline by stratifying for highest quartile of opioid (Z49.3
mean daily morphine equivalents). All analyses were performed using IBM SPSS software, version 21 (IBM SPSS, Armonk, New York), SAS software, version 9.3 (SAS Institute, Cary, North Carolina), and Stata software, version 12 (Stata Corp, College Station, Texas).
RESULTS Sixty-six nursing home patients enrolled in the study. Mean age was 74 years, 61% were male, and 36% were functionally independent or needed only supervision for their activities of daily living. Mean duration of nursing home stay before enrollment was 2.1 years. Enrolled patients received a mean of 8 non-
Table I. Characteristics of patients.
Characteristic
Nursing Home
Nursing Home Patients
Patients Prescribed Opioids
Prescribed Opioids Who Experienced Acute Illness
Patients, n Age at enrollment, y, mean (SD) [range] Sex, male, n (%)
66
45
73.9 (11.9) [4696] 40 (60.6)
75.1 (12.2) [5395] 25 (55.6)
Race, n (%) White Black or African American
39 (59.1) 22 (33.3)
30 (66.7) 13 (28.9)
Unknown or not reported
4 (6.1)
2 (4.4)
58 (87.9)
42 (93.3)
Ethnicity, n (%) Not Hispanic or Latino Hispanic or Latino Unknown Cognitive function (MDS-CPS) Intact or Borderline Intact (01) Mild or moderate impairment (23) Moderate severe to severe (45)
7 (10.6)
2 (4.4)
1 (1.5)
1 (2.2)
51 (77.3)
35 (77.8)
7 (10.6) 5 (7.5)
6 (13.3) 3 (6.6)
24 (36.7) 15 (22.7)
15 (33.3) 11 (24.3)
Physical impairment (ADL) Independent or supervision (01) Limited or extensive (23) Dependent or total (46) Duration of NH stay, d, mean (range) No. of nonopioid medications, mean (SD)
24 (36.3) 759.3 (134895) 7.76 (3.12)
19 (42.2) 752.09 (204895) 7.82 (3.03)
No. of chronic medical conditions, mean (SD)
3.71 (1.65)
3.91 (1.68)
Hospital days in previous year, mean (SD) No. of illnesses
19.0 (34.8)
19.16 (37.80) 104
Illness severity score (IPFS), mean (SD) [range 039; higher ¼ worse]
2.68 (4.06) [015]
Hospital admission, n (%) If admitted, days of hospital stay, mean (SD) [range]
24 (23) 8.38 (7.85) [135]
ADL ¼ activities of daily living; CPS ¼ Cognitive Performance Scale; IPFS ¼ Inpatient Physiologic Failure Score; MDS ¼ Minimum Data Set; NH ¼ nursing home.
1558
Volume 36 Number 11
S.E. Redding et al. opioid medications, and had a mean of 3.7 chronic conditions. Of the 66 patients receiving opioids that were followed for a mean of 10.9 months, 45 experienced a total of 104 acute illnesses. The majority (77%) of these acute illnesses were treated in the nursing home. The most common diagnoses were cellulitis, urinary tract infection, upper and lower respiratory infections, chronic obstructive pulmonary disease exacerbation, sepsis, and dehydration. Characteristics of the subgroup that became acutely ill during the study period were similar to those of the group as a whole (Table I). During 64 illnesses (62%), patients experienced any reduction in their opioid medication dosage. The mean dose reduction was 63.9%, from 63 daily morphine milligram equivalents to 29. During 39 illnesses (38%), patients experienced a significant opioid interruption (complete discontinuation or >50% dose reduction). Significant interruptions occurred on average on day 5 of acute illness (mean [SD] days, 5.06 [4.02]) and lasted for a mean of 3 days (range 118 days) (Table II). The most frequently interrupted medication was oxycodone (50% of all interruptions). Interruptions occurred most frequently in the nursing home (63%), and the
Table II. Characteristics of opioid interruptions. Episodes of illness among patients 104 prescribed opioids, n Episodes with opioid interruption, n (%) 64 (62) Days from illness onset to interruption, mean (SD) 5.06 (4.02) Duration of interruption, mean days (range) 3 (118) Complete discontinuation, n (%) 22 (21) Daily morphine milligram equivalents, mean (SD) Before interruption 62.97 (87.50) During interruption 29.12 (48.55) % Dose reduction, mean (SD) 63.9 (29.9) Interruption location, n (%) In nursing home (without or before 40 (63) hospital transfer) In hospital 14 (22) In nursing home after hospital discharge 9 (14) Interrupted medication, n Oxycodone 32 Methadone 5 Hydrocodone 3 Codeine 1 Hydromorphone 1 Multiple opioids 20
November 2014
remainder occurred in hospital (22%) or after readmission to the nursing home from the hospital (14%). In bivariate analysis, increasing age was found to be associated with a lower risk of opioid interruption (odds ratio [OR] ¼ 0.947; 95% CI, 0.8980.999; P ¼ 0.04). In a multivariable analysis, only residence at one nursing home (the VA site) was associated with a lower risk of interruption (OR ¼ 0.073; 95% CI, 0.0090.597; P o 0.015) (Table III). The mean baseline pain score was 5.9. During illnesses in which patients experienced a significant interruption in opioid dosing, there were insignificant changes in mean (SD) pain score (difference 0.50 [2.66]; 95% CI, 3.16 to 2.16) and withdrawal score (difference 0.91 [3.12]; 95% Cl, 4.03 to 2.21) within those patients as compared with preinterruption. Similarly, during illnesses in which patients experienced no significant reduction in opioid dosing, there were no significant changes in mean (SD) pain score (difference 0.96 [2.69]; 95% CI, 3.65 to 1.74) or withdrawal score (difference 0.34 [2.16]; 95% CI, 1.82 to 2.5) within those patients before and after day 5 of acute illness (Figures 1 and 2). However, when compared with patients without interruptions, patients with interruptions had larger increases in pain scores over time (difference 0.09 points per day; 95% CI, 0.01 to 0.019; P = 0.08). In particular, among patients who received the highest quartile of opioid dose, those with interruptions experienced increases in pain scores over time that were 0.22 points per day larger (95% CI, 0.02 to 0.41; P=0.03) than those without interruption. Withdrawal scores were not associated with drug interruption over time (point estimate 0.12; 95% CI, 0.07 to 0.30; P = 0.21), regardless of opioid dose before interruption. Of 18 charts reviewed to ascertain the reasons for opioid interruption, 7 (39%) had clear documentation and a clinical reason for opioid interruption, the most common being altered mental status. On the other hand, 11 (61%) had no documentation for the reason for interruption. In many of these cases, the interruption occurred during a transition of care such as change in staffing, transfer to hospital, or transfer to a different unit within the nursing home.
DISCUSSION In this study, we found that a large proportion of nursing home residents experienced reductions in
1559
Clinical Therapeutics
Table III. Factors associated with opioid interruption. Independent Variable Patient characteristics Age Chronic medical conditions Physical function Cognitive function No. medications Opioid dose (morphine milligram equivalents) Nursing home site Illness characteristics Illness severity Hospitalization (yes or no) Increased depression –– Increased pain Change in ability to communicate *
Bivariate P Value
Multivariate P Value
0.040* 0.911 0.559 0.821 0.407 0.133
0.304 0.654 0.826 0.694 0.127 0.110
0.963 0.896 0.947 0.889 1.346 1.010
0.440
0.015*
0.073 (0.0090.597)
0.274 0.180 0.898 –– 0.913 0.547
Multivariate Odds of Interruption (95% CI)
0.188 0.116 0.999 –– 0.683 0.309
(0.8971.034) (0.5531.450) (0.5841.537) (0.4951.597) (0.9191.970) (0.9981.023)
0.813 (0.5981.106) 0.082 (0.0041.854) –– –– (15.242) 0.719 (0.1473.509) 0.385 (0.0612.422)
P r .05
12 10 8 Mean BPI Scores
6 4 2 0
–10
–5
5
10
15
20
Time of Interruption –2 –4
Time (days)
Figure 1. Mean pain scores (Brief Pain Inventory [BPI]) versus time (days) after interruption. Squares and dashed lines ¼ group with significant opioid interruption. Triangles with filled line ¼ group with no significant opioid interruption.
1560
Volume 36 Number 11
S.E. Redding et al.
12
10
Mean COWS Score
8
6
4
2
–5
0 5 Time of Interruption
10
15
20
–2 Time (days)
Figure 2. Mean withdrawal scores (Clinical Opiate Withdrawal Scale [COWS]) versus time (days) after interruption. Squares and dashed lines ¼ group with significant opioid interruption. Triangles with filled line ¼ group with no significant opioid interruption.
opioid prescription dosage during acute illness (62% of illnesses), and many of these events were complete discontinuations or large (>50%) dose reductions. These interruptions occurred regardless of whether the patient was transferred to a hospital. We found that during these interruptions, there were small, insignificant changes in pain scores or withdrawal symptoms within patients when compared with patients’ preinterruption scores. However, when compared with patients who did not have interruptions, those with significant interruptions had larger pain increases over time. The difference in pain trajectories was particularly significant among those receiving higher preinterruption doses of opioid medications. Among these patients, those with opioid interruption had pain that increased by 0.22 points per day more than those without interruption, which, during the course of a mean follow-up of 7 days, could result in pain that is 1.5 points (0.22 7) higher on a 10-point scale. This could represent a difference between experience of mild and moderate pain, or between moderate and severe pain, a clinically significant difference for patients. In addition, for US nursing facilities, this could impact the Centers for Medicare
November 2014
and Medicaid Services publically reported quality measure “percentage of residents with moderate-tosevere pain.” Our finding that there is a high rate of opioid interruption is consistent with earlier work that has examined medication interruption patterns during transfer from nursing homes to the hospitals,20 although rather surprisingly, whether a resident was transferred to the hospital or not was not associated with interruptions. This suggests that, in many cases, interruptions are driven by clinical factors and changes in management within the nursing home itself. Our finding that patients with large interruptions had higher pain scores than those without interruption is consistent with our hypothesis that interruptions in opioid medications can lead to worse pain control, an association that was even more prominent among those receiving higher doses of opioid medications in our study. We did not find that patients had worse withdrawal symptoms with large interruptions, which could be due to the difficulty of detecting changes in autonomic sympathetic activity in a study population that had high rates of chronic disease, medication use, and disability.
1561
Clinical Therapeutics We also found that among factors tested for their association with interruptions, only the nursing home site (the VA site) was associated with less interruption. Of note, patients at the VA nursing home, if needed, were admitted to the VA hospital in the same facility, which shared a similar electronic medical record and was served by the same pharmacy. There might be factors associated with the VA site of care, such as a shared electronic medical record, that may be associated with a lower risk of interruption. This is important because our review for reasons for interruptions found that the majority of cases did not have a documented clinical reason, and may have been precipitated by handoffs in care. Our finding is consistent with other studies on medication omissions in hospitals21 that highlight that clinicians need to improve their documenting of reasons for interruptions and to minimize unnecessary interruptions. There were several limitations to this study. Although our study had a structured protocol for monitoring for acute illnesses, it is possible that short 1- or 2-day delays in identifying acute illnesses could have occurred. Additionally, outcome assessments were performed every other day; therefore, it might have been possible to miss pain or withdrawal symptoms that occurred between assessments. We anticipate that the effects of these off days would be minimal, given that most interruptions occurred at a mean of 5 days after onset of the illness. Also, our power to observe smaller associations was limited by the small size of the cohort. The small sample size was in part a result of a relatively low prevalence of patients receiving opioid analgesics in the study setting, challenges with obtaining informed consent in the nursing home, and a lower than expected incidence of acute illness. Finally, we have missing data on a small proportion of patients (3%) that had hospitalizations at nonstudy hospitals.
CONCLUSIONS Interruptions in opioid medications occur frequently among nursing home patients during acute illness episodes. Clinicians should be wary of the possibility of worsened pain during acute illness if opioid medications are interrupted. Careful medication reconciliation is essential and clinicians need to consider the reasons for and the risks of opioid interruption in this setting.
1562
ACKNOWLEDGMENTS The authors are grateful to Daniel Signor for his role as database and statistical programmer and to Jessica Singleton, Jennifer Kwak, and Julia Siegel Breton, MD for their role in data collection. Dr. Redding contributed to data collection, data analysis, data interpretation, and manuscript writing; Dr. Liu contributed to study design, data collection and data analysis; Dr. Hung contributed to data analysis, data interpretation, and manuscript writing; Dr. Boockvar contributed to study design, data collection, analysis and interpretation, and manuscript writing.
CONFLICTS OF INTEREST This study was supported by Veterans Affairs Health Services Research and Development Service grants RCD 03-027-1 and REA 08-260. The study sponsor had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. One author (K. Boockvar) is also supported by the Greenwall Foundation. The contents do not represent the views of the US Department of Veterans Affairs or the United States Government.
REFERENCES 1. Won AB, Lapane KL, Vallow S, et al. Persistent nonmalignant pain and analgesic prescribing patterns in elderly nursing home residents. J Am Geriatr Soc. 2004;52:867–874. 2. Boockvar KS, Gruber-Baldini AL, Burton L, et al. Outcomes of infection in nursing home residents with and without early hospital transfer. J Am Geriatr Soc. 2005;53:590–596. 3. Suchanek M, Boockvar KS, Morrison RS, Fried T. Discontinuation of standing opioid orders in individuals transferred form nursing home to hospital. J Am Geriatr Soc. 2006;54:S204–S229. 4. Jaffe JH. Misinformation: euphoria and addiction. In: Hill CS Jr, Fields WS, eds. Advances in Pain Research and Therapy, Vol 11. New York: Raven Press; 1989:163–173. 5. Kosten TR, O’Conner PG. Management of Drug and Alcohol Withdrawal. New Engl J Med. 2003;348:1786– 1795. 6. Cowan DT, Wilson-Barnett J, Griffiths P, et al. A randomized, double-blind, placebo-controlled, cross-over pilot study to assess the effects of long-term opioid drug consumption and subsequent abstinence in chronic
Volume 36 Number 11
S.E. Redding et al.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
noncancer pain patients receiving controlled-release morphine. Pain Med. 2005;6:113–121. Gerety MB, Cornell JE, Plichta DT, Eimer M. Adverse Events Related to Drugs and Drug Withdrawal in Nursing Home Residents. J Am Geriatr Soc. 1993;41:1326–1332. Hung WW, Liu S, Boockvar KS. A prospective study of symptoms, function, and medication use during acute illness in nursing home residents: Design, rationale and cohort description. BMC Geriatr. 2010;10:47. Morris JN, Fries BE, Morris SA. Scaling ADLs within the MDS. J Gerontol A Biol Sci Med Sci. 1999;54: M546–M553. Morris JN, Fries BE, Mehr DR, et al. MDS Cognitive Performance Scale. J Gerontol. 1994;49:M174– M182. Protocols for Physician Notification. In: Levenson S, Vance J, eds. Assessing Patients and Collecting Data on Nursing Facility Patients. Columbia, MD: American Medical Directors Association; 2004. Gray LK, Smyth KA, Palmer RM, et al. Heterogeneity in Older People: Examining Physiologic Failure, Age, and Comorbidity. J Am Geriatr Soc. 2002;50:1955–1961. Mendoza T, Mayne T, Rublee D, Cleeland C. Reliability and validity of a modified Brief Pain Inventory short form in patients with osteoarthritis. Eur J Pain. 2006;10: 353, e361. Wesson DR, Ling W. The Clinical Opiate Withdrawal Scale (COWS). J Psychoactive Drugs. 2003;35:253–259. Pautex S, Herrmann F, Le Lous P, et al. Feasibility and Reliability of Four Pain Self-Assessment Scales and Correlation with an Observational Rating Scale in Hospitalized Elderly Demented Patients. J Gerontol A Biol Sci Med Sci. 2005;60: 524–529. AGS Panel on Chronic Pain in Older Persons. The management
November 2014
17.
18.
19.
20.
21.
of chronic pain in older persons. American Geriatrics Society. J Am Geriatr Soc. 1988;46:635–651. Patanwala AE, Duby J, Waters D, Erstad BL. Opioid conversions in acute care. Ann Pharmacother. 2007; 41:255–266. Ripamonti C, Groff L, Brunelli C, et al. Switching from morphine to oral methadone in treating cancer pain: what is the equianalgesic dose ratio? J Clin Oncol. 1998; 15:3216–3221. Naranjo CA, Busto U, Sellers EM, et al. A method for estimating the probability of adverse reactions. Clin Pharmacol Ther. 1981;30:239– 245. Boockvar K, Fishman E, Kyriacou CK, et al. Adverse events due to discontinuations in drug use and dose changes in patients transferred between acute and longterm care facilities. Arch Intern Med. 2004;164:545–550. Warne S, Endacott R, Ryan H, et al. Non-therapeutic omission of medications in acutely ill patients. Nurs Crit Care. 2010;15:112– 117.
Address correspondence to: Kenneth Boockvar, MD, James J Peters VA Medical Center, 130 West Kingsbridge Road, Bronx, New York 10468. E-mail: [email protected]
1563
