Article

Intravenous opioid dosing and outcomes in emergency patients: a prospective cohort analysis

Original Contribution

Intravenous opioid dosing and outcomes in emergency patients: a prospective cohort analysis

Alec B. O’Connor MD, MPH a,?, Frank L. Zwemer MD, MBA b,1,

Daniel P. Hays PharmD b,c,2, Changyong Feng PhD d

aDepartment of Internal Medicine, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA bDepartment of Emergency Medicine, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA cDepartment of Pharmacy, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA dDepartment of Biostatistics and Computational Biology, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA

Received 28 May 2009; revised 23 June 2009; accepted 24 June 2009

Abstract

Objectives: Pain management in emergency department (ED) patients is variable and often inadequate. This study sought to (1) describe the variability in intravenous opioid dosing and (2) compare the outcomes that result from the most commonly prescribed opioid doses.

Methods: This prospective cohort study enrolled emergency patients who were prescribed Intravenous morphine or hydromorphone as their initial analgesic. Subjects were interviewed at the time of opioid administration and 1 to 2 hours after opioid administration. Outcomes included the numeric pain score change (using a 0-10 scale), the proportion achieving a 50% pain score reduction, and the proportion developing side effects. Logistic regression was used to assess the effects of demographic, clinical, and treatment variables on outcomes.

Results: Six hundred ninety-one patients were analyzed. Initial equianalgesic dosages varied by a factor of 27 (from 1 mg morphine to 4 mg hydromorphone). Opioid dose titration occurred in only 21% of patients.

Outcomes were similar across the range of opioid dosages before and after adjusting for potentially confounding variables. Among patients not taking opioids at home who received a total of 4 mg of morphine or less. 48% achieved at least a 50% pain score reduction and 60% did not want additional analgesics.

Conclusions: We found marked opioid dosing variability and infrequent opioid dose titration. A substantial number of ED patients with severe pain responded well to relatively low opioid dosages. Improved ability to predict opioid dose requirements and strategies that increase the use of opioid dose titration in ED patients are needed.

(C) 2010

Results from this study have been presented in abstract form at the annual meetings of the American Academy of pain medicine (New Orleans, La, February 2007) and the Society of Academic Emergency Medicine (Chicago, Ill, May 2007).

* Corresponding author. Box MED/HMD, Strong Memorial Hospital, Rochester, NY 14642, USA. Tel.: +1 585 275 4912; fax: +1 585 276 2144.

E-mail address: [email protected] (A.B. O’Connor).

0735-6757/$ – see front matter (C) 2010 doi:10.1016/j.ajem.2009.06.009

Introduction

Pain is one of the most common reasons for emergency department (ED) visits, yet inadequate pain relief continues to be commonplace [1-4]. A number of studies have described poor analgesic care of ED patients with severe pain, including failure to administer analgesics to many patients with painful diagnoses, substantial delays before analgesics are administered, and relatively infrequent use of opioids [3-9]. Comparatively little is known about the clinical outcomes associated with opioid analgesic adminis- tration to patients in the ED.

In theory, higher opioid dosages result in both greater analgesia and greater side effect burden. Opioid-naive individuals with uniform surgical injuries require widely variable opioid dosages for good analgesia, and trial and error is required to determine an individual’s dose require- ment [10]. For this reason, individualized opioid dose titration based both on analgesic response and on side effect tolerance has been recommended for patients with severe pain [10]. However, because resources are limited in EDs, a reasonable goal is to optimize the initial opioid bolus dose to reduce the need for further opioid dose titration.

The optimal initial opioid dose for ED patients with severe pain has not been determined. We recently reported that only 43% of patients in an ED cohort who received Intravenous opioids achieved a 50% pain score improvement after receiving a median of 4 mg morphine (0.055 mg/kg) [11]. The dosages in our cohort were quite variable but on average were approximately half of a dose (0.1 mg/kg) that has previously been found to be insufficient to relieve pain in ED patients [12] and one third of the dosages administered via aggressive opioid titration to ED patients (0.15 mg/kg)

[13] and postanesthesia care unit patients (0.15 mg/kg) [14]. The outcomes resulting from the lower dosages commonly used in practice have not been determined.

Goals of this investigation

In this analysis of an ED cohort of patients who received intravenous opioids, we aim to (1) describe intravenous opioid prescribing behavior in this population and (2) compare the outcomes resulting from opioid dosing variability. We hypothesized that patients receiving higher equiAnalgesic doses would achieve substantially better analgesia than patients receiving lower equianalgesic doses after adjusting for measurable covariates.

Materials and methods

Study setting and population

Patients were prospectively enrolled in this observational cohort study between July 2004 and November 2006. The

study occurred in an Urban academic ED with 93 000 annual visits. Most patients served are non-Hispanic whites, with 17% of patients self-pay or Medicaid. There are no specific opioid dosing protocols used by providers in the ED. The study was approved and monitored by the university’s institutional review board.

Emergency department patients at least 18 years old were eligible for inclusion if intravenous morphine or hydro- morphone was administered as their initial analgesic. Exclusion criteria included prior or simultaneous adminis- tration of any analgesic medications in the ED or via emergency medical services (to focus on the effects of the opioids), mental capacity inadequate to provide informed consent, inability to communicate in English, or unstable vital signs. Patients were enrolled before administration of the initial opioid dose when possible (ie, immediately after the opioid order), but enrollment was allowed at any time within 2 hours of the administration of the initial opioid dose (we previously reported similar results for patients whose initial pain score was determined before vs after initial opioid administration) [11].

Study protocol

All patient screening, consenting, and data collection was performed by trained ED enrollers [15] who staffed the ED continuously (with few exceptions) from 8:00 AM to midnight. Potential study subjects were identified by screening triage diagnoses for potentially painful diagnoses and by provider and nursing notification about new orders for intravenous morphine or hydromorphone. All enrolled subjects provided informed consent after an institutional review board-approved protocol.

The ED enrollers sequentially interviewed subjects

(preanalgesic assessment), then providers, then subjects (postanalgesic assessment between 1 and 2 hours after the administration of the initial dose), and finally the subject’s nurse. In addition, each subject’s chart was reviewed after hospital discharge. Additional detail describing the protocol and study measures can be found elsewhere [11].

Study measures

To compare morphine and hydromorphone doses, hydro- morphone doses were converted into equivalent milligrams of morphine (Mmg) using a standard equianalgesic dosing ratio (10 mg of IV morphine = 1.5 mg of Intravenous hydromorphone) [10,16,17].

In addition to collecting standard clinical and demo- graphic information about subjects, we asked providers to quantify their suspicion for patient “drug-seeking behavior (feigned symptoms to obtain narcotics)” using a 1 to 5 Likert scale with anchors of 1 = “not concerned” and 5 = “very concerned.” Although this scale is unvalidated, we were unable to find a validated scale to measure concern for drug seeking. In the analyses, we considered providers to believe

patients to be possibly drug-seeking if the providers’ suspicion for drug-seeking was rated as 2 or higher on this scale because this represented the highest quintile of scores. We measured clinical outcomes in several different ways.

We asked patients to quantify their Pain severity using the standard verbal 0 to 10 scale (anchors of 0 = “no pain,” 10 = “worst pain possible”) before analgesic administration and again at the postanalgesic survey. A 50% pain score reduction was used to indicate a substantial analgesic response because it correlates with patient reports of “treatment success” and with being “very much improved” [18-20]. Subjects and nurses were independently asked about the development of any “side effects” or “new symptoms” after opioid adminis- tration; side effects were considered to be present if either the patient or nurse reported the presence of “side effects” or “new symptoms” or if a side effect was documented in the chart. A single investigator (ABO) classified all reported side effects while blinded to drug and dose. Additional outcome measures included a categorical pain relief scale (from 1 = “no relief” to 5 = “complete relief”) [21], a categorical pain treatment satisfaction scale (from 1 = “very dissatisfied” to 6

= “very satisfied”) [22], desire for additional pain medication at the postanalgesic interview, and occurrence of severe side effects, defined as developing any of the following: unstable vital signs (systolic blood pressure [SBP] b 90 or respiratory rate [RR] b 10), reduced oxygen saturation (N10% change to below 90% or need for increased oxygen flow rate), or an order for naloxone after administration of morphine or hydromorphone.

Data analysis

For this analysis of opioid dosing, a priori we defined the comparison between the 4 most commonly prescribed opioid dosages to be the primary analysis, and the pain score change (preanalgesic score minus postanalgesic score), the proportion achieving a 50% pain score improvement,

and the proportion developing side effects to be the primary outcome variables.

All statistical analyses were performed using SAS software, version 9.1 (SAS Institute Inc, Cary, NC). We attempted to adjust the pain score change using linear regression (the PROC GLM procedure in SAS) but found the covariates to be almost colinear and the matrix singular, even after stratifying the dependent variable and limiting the number of covariates. Therefore, we compared pain score changes within subgroups defined by common demographic and clinical variables. The 4 cohorts we compared were (1) the entire cohort (“all patients”); (2) the subset of patients not taking any home opioids within 48 hours of presentation (the “no home opioids” cohort), which was part of the prespecified data analysis plan; (3) a restricted subset created by excluding patients with characteristics we found to be associated with the likelihood of achieving a 50% pain score change in our previous analysis of this cohort using logistic regression [11]; specifically, this cohort (called the “restricted subset”) excluded from the whole cohort patients with any of the following characteristics: (a) home use of long-acting opioids, (b) prescriber concern for drug-seeking behavior, (c) additional analgesic administration, or (d) age N65. The fourth cohort we analyzed was the “most restricted subset,” which consisted of the “restricted subset” patients with an initial pain score of 10 (the mode) because our previous analysis found a strong association between initial pain score and a final pain score of 7 or greater [11].

Odds ratios for the association between dosage and (1) 50% pain score reduction and (2) the development of side effects were adjusted via logistic regression using the PROC LOGISTIC procedure in SAS. The following variables were included in the regression models, based on being major demographic variables or because they were distributed differently between the different opioid dosages: age, sex, weight, home opioid use, long-acting opioid use, pain diagnosis (the 4 most common diagnoses [abdominal pain,

Fig. 1 Study enrollment.

Table 1 Comparison of the demographic, clinical, and treatment characteristics by opioid dose for the entire (all patients) cohort

Morphine 2 mg

Morphine 4 mg

Hydromorphone 1 mg

Hydromorphone 2 mg

Demographics

n = 131

n = 291

n = 121

n = 64

Age, y (median [IQR])

42 (26-58)

40 (28-53)

40 (30-52)

38 (31-48)

Female (n [%])

88 (67)

175 (60)

73 (60)

36 (56)

Race

White, non-Hispanic (n [%])

83 (65)

191 (66)

82 (68)

37 (62)

African American (n [%])

32 (25)

76 (26)

27 (22)

19 (32)

Hispanic (n [%])

6 (5)

15 (5)

9 (7)

3 (5)

Clinical characteristics

Initial pain score,

8.0 (7.0-10)

9.0 (7.0-10)

10 (8.0-10)

10 (9.0-10) ?

0-10 scale (median [IQR])

Any home opioid use (n [%])

32 (24)

68 (23)

50 (41)

35 (55) ?

Long-acting opioid use (n [%])

4 (3)

7 (2)

16 (13)

6 (9) ?

History of opioid allergy (n [%])

14 (11)

40 (14)

25 (21)

21 (33) ?

History of substance abuse (n [%])

13 (10)

24 (8)

10 (8)

9 (14)

History of chronic medical

36 (27)

60 (21)

24 (20)

22 (34)

disease (n [%])

Pain diagnosis

Abdominal pain (n [%])

63 (48)

134 (46)

43 (36)

23 (36)

Trauma (n [%])

16 (12)

35 (12)

12 (10)

5 (8)

Acute fracture (n [%])

12 (9)

20 (7)

5 (4)

3 (5)

Back pain (n [%])

6 (5)

18 (6)

15 (12)

15 (23) ?

Kidney stone (n [%])

5 (4)

11 (4)

19 (16)

5 (8) ?

Provider concern for

35 (27)

85 (29)

33 (27)

17 (27)

patient stability (n [%])

Provider concern for

21 (16)

36 (12)

30 (25)

28 (44) ?

drug-seeking (n [%])

Treatment characteristics

Initial equianalgesic dose, Mmg

2.0

4.0

6.7

13 ?

Weight-adjusted initial

0.027 (0.022-0.031)

0.049 (0.043-0.061)

0.081 (0.068-0.10)

0.16 (0.13-0.19) ?

equianalgesic dose,

Mmg/kg (median [IQR])

Total weight-adjusted

0.028 (0.024-0.033)

0.055 (0.044-0.068)

0.092 (0.073-0.13)

0.17 (0.14-0.24) ?

equianalgesic dose,

Mmg/kg (median [IQR])

Coadministered antiemetic (n [%])

49 (37)

128 (44)

64 (53)

35 (55) ?

Additional analgesic

25 (19)

61 (21)

36 (30)

22 (34) ?

administered before

postanalgesic survey (n [%])

Provider goal of full

71 (54)

156 (54)

65 (54)

30 (47)

pain relief (n [%])

Analgesic outcomes

Postanalgesic pain score,

5.0 (3.0-7.5)

5.0 (3.0-7.0)

5.0 (2.0-8.0)

6.0 (3.0-8.0)

0-10 scale (median [IQR])

Change in pain score,

3.0 (1.3-5.0)

3.0 (1.0-5.0)

4.0 (1.8-6.0)

3.0 (1.0-5.0)

0-10 scale (median [IQR])

>=50% Pain score reduction (n [%])

56 (43)

127 (44)

59 (49)

22 (34)

>=30% Pain score reduction (n [%])

73 (56)

171 (59)

77 (64)

37 (58)

Desire for additional analgesics (n [%])

49 (37)

136 (47)

62 (51)

41 (64) +

Pain relief, 1-5 scale (median [IQR])

3.0 (2.0-4.0)

3.0 (2.0-4.0)

3.0 (2.0-4.0)

3.0 (2.0-4.0)

Satisfaction with pain treatment,

5.0 (4.0-5.5)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (3.0-5.5)

1-6 scale (median [IQR])

Side effects

Serious side effects (n [%])

1 (1)

3 (1)

0

0

Any side effects (n [%])

20 (15)

67 (23)

34 (28)

17 (27)

Central nervous system

7 (5)

39 (13)

21 (17)

8 (13) ?

side effects (n [%])

Table 1 (continued)

Morphine 2 mg

Morphine 4 mg

Hydromorphone

Hydromorphone

1 mg

2 mg

Sedation (n [%])

5 (4)

26 (9)

18 (15)

7 (11) ?

Dizziness (n [%])

4 (3)

11 (4)

0

3 (5)

Confusion (n [%])

0

7 (2)

5 (4)

1 (2)

Nausea (n [%])

5 (4)

17 (6)

9 (7)

3 (5)

Pruritis (n [%])

1 (1)

1 (0)

3 (2)

3 (5) ?

Dry mouth (n [%])

0

4 (1)

4 (3)

2 (3)

Headache (n [%])

0

3 (1)

0

1 (2)

Hypotension (n [%])

1 (1)

3 (1)

0

0

Disposition

Admitted (n [%])

33 (25)

83 (29)

32 (26)

17 (27)

Length of stay,

0.40 (0.30-1.9)

0.50 (0.30-2.0)

0.42 (0.20-2.0)

0.50 (0.30-2.0)

d (median [IQR])

trauma, back pain, and kidney stone pain] were included in the models), history of opioid allergy, concern for drug- seeking behavior, history of chronic illness, initial pain score, time between dose and postanalgesic survey, coadministra- tion of antiemetics, and administration of additional analge- sics. We removed variables from the models if they were not associated with the outcome variable (P N .2). The goodness of fit for the logistic regression models was assessed using the Akaike information criterion. The logistic regression models for the proportion achieving a 50% pain score reduction had excellent fitting, whereas the models for side effects with covariates had Akaike information criterion values of at least

* P <= .05.

+ P b .01.

? P b .001.

1.6 lower than the models without covariates.

Many of the continuous variables were not normally distributed. For this reason, we used the Kruskal-Wallis test for hypothesis testing involving continuous variables. We used the Fisher exact test for the hypothesis testing of categorical variables. A priori we set our threshold for statistical significance to be P b .05.

Sensitivity analyses

We considered home opioid use to be a major potential confounder because it may affect both dosing (providers may give higher opioid doses to patients taking home opioids) and outcomes (patients taking home opioids may have less pain relief and side effects from a given dosage due to opioid tolerance). As described above, our primary analyses adjusted for both home opioid use and long-acting opioid use (these are typically prescribed at higher dosages and over more prolonged time, making significant opioid tolerance more likely). To exclude the possibility that our findings could be explained by inadequately adjusting for home opioid use, we also analyzed our data after removing all subjects taking home opioids (the “no home opioids” cohort).

Because we did not find a clear analgesic dose

response effect in our primary analysis (ie, higher dosages

were not clearly associated with greater pain score reductions), we also analyzed our data by stratifying the cohort into quartiles based on weight-adjusted equianal- gesic dose (Mmg/kg).

Results

Description of the cohort, treatment, and outcomes

A total of 691 patients were included in the analysis (Fig. 1), of whom 60% were female. As described previously, patients had a median age of 40 (interquartile range [IQR], 29-52) and an initial pain score of 9 (IQR, 8-10). Two hundred twenty (32%) were taking home opioids [11].

Initial opioid dosages ranged from 1 mg morphine to 4 mg hydromorphone (equivalent to 27 mg of morphine, or “Mmg”), with a median equianalgesic dosage of 4.0 Mmg (IQR, 4.0-6.7) or 0.055 Mmg/kg (IQR, 0.038-0.079).

Additional analgesics (both opioid and non-opioid) were administered to 24% of the cohort (169/691) before the postanalgesic survey, which occurred a median of 87 minutes after the initial opioid dose was administered: 146 (21% of total) received at least one additional intravenous opioid dose, 21 (3% of total) received 2 additional intravenous opioid doses, and 22 (3% of total) received intravenous ketorolac as an additional analgesic. Of patients with a final pain score of 7 or higher, 76/249 (31%) received an additional intravenous opioid dosage. Most patients who received an additional intravenous opioid dose received the same opioid (75% of morphine patients and 100% of hydromorphone patients), and most received the same dose as the original dosage.

The most commonly administered dosages were 4 mg morphine (n = 291), 2 mg morphine (n = 131), 1 mg hydromorphone (n = 121), and 2 mg hydromorphone

(n = 64), which together accounted for 89% of the initial dosages. There were no significant differences in the demographic characteristics of the patients who received different dosages (Table 1).

Compared to patients who received morphine, patients who received hydromorphone had higher initial pain scores and were more likely to be taking home opioids, be considered possibly drug-seeking, have an opioid allergy, have a diagnosis of either back pain or kidney stone, receive coadministered antiemetics, and receive additional analgesics (Table 1).

Association between opioid dosages and analgesic outcomes

The distributions of pain score changes associated with the most commonly administered opioid dosages were similar (Appendix Fig. 1). Opioid dosage was not associated with the pain score change, the likelihood of achieving a 50% pain score improvement, or the likelihood of developing side effects (Table 1). Other Treatment outcomes were similar for the different dosages except that patients who received 2 mg hydromorphone were more likely to want additional analgesics and develop pruritis, and patients who received 2 mg morphine were less likely to develop central nervous system side effects, especially sedation (Table 1).

Comparisons of the median pain score changes associated with the most commonly administered dosages for different subsets of the cohort are shown in Fig. 2. In the restricted subset, the differences in median pain score change were statistically significant (P = .04) and there was a dose response trend for 2 mg morphine, 4 mg morphine, and 1 mg hydromorphone. However, this finding was not stable as we further restricted the subset to those patients with an initial pain score of 10 (the mode) (Fig. 2).

In the logistic regression analyses, there was no associa- tion between opioid dosage and the likelihood of achieving a 50% pain score reduction (Table 2, all patients). There was a trend toward greater likelihood of side effects as opioid dosage increased, but the differences were not statistically significant (Table 2).

Sensitivity analyses

      1. Patients not taking opioids at home

The results of the analyses after excluding the 220 patients who were taking home opioids (the no home opioids cohort) were not substantially different from the analyses of the entire cohort (Fig. 2, Table 2, no home opioids group; Appendix Table 1, Appendix Fig. 1). Sixty-two percent of the patients not taking home opioids (291/471) received a total of 4 Mmg or less; 48% of these patients (139/291) achieved at least a 50% pain score reduction and 60% (174/ 291) did not want additional analgesics.

Fig. 2 Box plots demonstrating the pain score improvement by opioid dosage. The boxes show the interquartile range around the median line. The surrounding bars demonstrate the 10th and 90th percentile values. All patients indicates the entire cohort; no home opioids, the cohort remaining after excluding the patients taking home opioids. The restricted subset consists of patients meeting each of the following criteria: no long-acting opioid use, not considered potentially drug-seeking, no additional analgesics, and age of less than 65 years. The most restricted subset is the restricted subset patients who had an initial pain score of 10 (the mode). P N .05 for all within subset comparisons except P = .04 for the comparisons in the restricted subset.

      1. Outcomes associated with weight-adjusted equianalgesic dose

Adjusted odds ratio (95% CI)

Reference

1.92 (1.01-3.65)

2.62 (1.22-5.59)

1.92 (1.01-3.65)

Because we failed to find a clear analgesic dose- response relationship in our primary analyses, we reanalyzed our data after stratifying the cohort into weight-adjusted equianalgesic dose quartiles (rather than stratifying by the most common doses) (Appendix Tables 2 and 3). There was no linear relationship between weight-adjusted dosage and pain score change (Appendix Fig. 2). As with our primary analyses, we did not find a consistent relationship between weight-adjusted equianal- gesic dosage and pain score change (Appendix Fig. 3). There was no association between dosage and 50% pain score reduction (Appendix Table 4). However, increasing dosages were associated with a trend toward increased risk of side effects in the cohort as a whole, and there was a statistically significant association between dose and side effect risk in the no home opioids cohort (Appendix Table 4).

No home opioids

>=50% Pain score reduction

Any side effects

Unadjusted odds ratio (95% CI)

Reference

1.94 (1.02-3.69)

2.73 (1.28-5.81) ?

Unadjusted odds ratio (95% CI)

Reference

1.23 (0.76-1.98)

1.65 (0.89-3.06)

Adjusted odds ratio (95% CI)

Reference

1.13 (0.68-1.86)

1.88 (0.97-3.66)

1.23 (0.76-1.98)

1.13 (0.68-1.86)

1.94 (1.02-3.69)

Morphine 2 mg is considered the reference. ‘All patients’ indicates the entire cohort; ‘no home opioids’, the cohort remaining after excluding the patients taking home opioids.

* P b .05 for comparison with morphine 2 mg.

Discussion

Table 2 Unadjusted and adjusted odds ratios of categorical outcomes associated with the most common dosages

We found substantial variability in the intravenous opioid doses administered to ED patients–initial opioid dosages varied by a factor of 27 (from 1 mg morphine to 4 mg hydromorphone, roughly equivalent to 27 mg morphine). Opioid dose titration occurred infrequently, even among patients who continued to have severe pain.

Adjusted odds ratio (95% CI)

Reference

1.60 (0.92-2.78)

2.35 (1.25-4.43)

2.22 (1.03-4.77)

We expected to find that patients receiving low opioid dosages would have inadequate pain relief compared to patients receiving higher dosages. We were particularly surprised to find similar pain score changes in patients receiving 2 mg morphine and 2 mg hydromorphone, which is roughly 7 times more potent, even after adjusting for measured potential confounders. Although we did not find improvED analgesia with higher equianalgesic dosages, we did find that side effects, particularly sedation, were more frequent.

Any side effects

Unadjusted odds ratio (95% CI)

Reference

1.66 (0.96-2.87)

2.17 (1.17-4.03) ?

Adjusted odds ratio (95% CI)

Reference

0.98 (0.63-1.52)

1.37 (0.80-2.34)

1.02 (0.52-2.02)

2.01 (0.97-4.17)

We suspect that we did not find improved analgesic outcomes with higher dosages because we were unable to measure some of the confounding variables affecting opioid prescribing. We did find several patient character- istics that were associated with preferential prescribing of hydromorphone, including higher initial pain score, home opioid use, history of opioid allergy, certain diagnoses (back pain and kidney stone pain), and provider concern for drug-seeking behavior. Hydromorphone is not known to provide benefits over morphine for any of these character- istics, but on average, hydromorphone was prescribed in substantially higher doses than morphine. Several of the patient characteristics associated with hydromorphone prescribing would be expected to increase opioid dose requirement, so it appears that providers frequently chose to prescribe hydromorphone rather than prescribe a higher dose of morphine.

Opioid dose

All patients

>=50% Pain score reduction

Unadjusted odds ratio (95% CI)

Reference

1.04 (0.68-1.57)

1.27 (0.78-2.09)

Morphine 2 mg

Morphine 4 mg Hydromorphone

  1. mg (6.7 Mmg) Hydromorphone
  2. mg (13 Mmg)

0.70 (0.38-1.31)

It is likely that there were other unmeasured confounders that caused providers to appropriately reduce the opioid dosage for some patients and appropriately increase the dosage for others. We previously reported the strong association between physician concern for possible drug-seeking behavior and poor analgesic outcomes in this cohort [11]. The present analysis demonstrates an association between concern for drug-seeking behavior and prescribing hydromorphone in relatively high equi- analgesic doses, which would be expected to provide greater analgesia. Two related confounders that we did not measure that might have contributed to our findings are physician impression of pain severity and physician comfort with the presence of a painful diagnosis [23]. For example, it has been shown that ED patients with a long bone fracture are twice as likely to receive opioids as patients with similar pain scores without a long bone fracture [6]. A different study found that factors associated with greater confidence in a specific diagnosis increased the likelihood that providers would prescribe an opioid [24]. Others have found that physician impression that patients were exaggerating pain symptoms was not associated with whether patients received pain treatment, but it was associated with smaller pain score improve- ment with treatment [25]. It is also possible that unmeasured confounders that may affect patient response to opioids were asymmetrically distributed between dosing groups.

The optimal opioid dosing strategy for ED patients has not been determined. Lvovschi and colleagues [13] recently described a cohort of 621 ED patients managed with a morphine titration strategy (2- or 3- mg boluses, depending on weight, administered every 5 minutes until pain relief), which produced excellent analgesia (mean final pain score of 27 down from initial pain score of 84 on a 0-100 pain scale) after a median of 3 dosages. However, concern about the availability of sufficient nursing resources to administer an aggressive opioid titration strategy in a busy ED has caused other investigators to seek “a compromise” strategy using a higher initial opioid bolus dose to reduce subsequent nursing requirements [26]. Recent randomized clinical trials have found comparable clinical outcomes between 0.1 mg/kg morphine and 0.015 mg/kg hydro- morphone [27] and between 0.1 mg/kg and 0.15 mg/kg intravenous morphine [28]. Our findings indicate that a relatively large percentage of patients respond well to lower dosages–more than one third of all patients not taking home opioids received less than 4 Mmg and did not want additional analgesics. Similarly, nearly one fourth of patients in Lvovschi’s cohort were titrated to a total dose of less than 6 mg of morphine [13].

Our data could be interpreted as supporting the use of the opioid dose titration strategy. Because the 2-mg morphine dosage was associated with a lower risk of side effects and produced clinically equivalent pain score changes, it may be the best initial dose (of those studied) for patients who are not

expected to require high opioid dosages (such as those taking significant home opioid dosages), provided that additional dosages are administered as needed. However, our data also showed that opioid dose titration is frequently not achieved in our ED.

Several recent studies have evaluated opioid dosing protocols for emergency patients [13,26,29-31], which seem to increase the likelihood of adequate and timely opioid administration. Given the improved analgesic outcomes associated with Combination therapy [10,32-34], acetami- nophen or a nonsteroidal anti-inflammatory agent or both should also be routinely administered to patients with severe pain. We believe that analgesic protocols that include both nonopioid analgesic administration and opioid dose titra- tion, using small opioid doses repeated as needed after the peak effect of intravenous opioid has occurred (eg, every 5- 15 minutes), deserve exploration. Our findings suggest that some patients will respond well to low initial opioid dosages. It may be that certain patients can be identified as having higher Opioid requirements (such as those taking home opioids [11]) and placed on a higher dose protocol. Similarly, future research that helps identify patients at the greatest risk of significant side effects may help to avoid overdosing these patients.

Limitations

This study has a number of limitations. Observational studies are prone to a variety of types of bias. As described above, we believe that there were unmeasured confounding variables that explain why we did not find better analgesic responses among the patients who received higher equia- nalgesic doses after we adjusted for measured potential confounders. There were also limitations in our ability to adjust for covariates, particularly in our attempts to adjust pain score changes using linear regression. One of the challenges we encountered was the large number of potential confounders and the low frequency of patients with some of the characteristics (eg, no specific diagnoses occurred with sufficient frequency to allow meaningful comparisons between patients who received different doses). Additional randomized clinical trials are needed to clarify the clinical outcomes resulting from different dosages under controlled conditions.

It is possible that results from our patient population, which had fairly specific inclusion requirements, may not apply to the more general ED patient population. Variation in opioid prescribing practices by hospital are known to exist [35], so prescribing patterns in our ED may not be generalizable to other EDs, although the average equianal- gesic doses administered to our study population are similar to those administered to patients in 2 EDs elsewhere [36]. The decision to prescribe intravenous opioids defined our population, and the threshold for administering opioids or administering them intravenously may vary from ED to ED. Finally, our comparisons between morphine and hydromor-

phone dosing are based on the most accepted equianalgesic conversion ratio (10 mg parenteral morphine = 1.5 mg parenteral hydromorphone) [10,16,17], although this ratio is not based on good data [37].

Conclusions

We found marked opioid dosing variability, infrequent opioid dose titration, and that higher initial dosages may be associated with greater side effect frequency, but we did not find that higher initial opioid dosages were associated with better pain relief. We suspect that this finding is explained by the presence of unmeasured cues that ED providers use to successfully predict opioid dose requirements. A significant proportion of this cohort achieved good pain relief despite receiving opioid doses that are considered small (2 or 4 mg morphine), which suggests that protocols giving larger initial bolus doses may overtreat a substantial number of patients. Improved ability to predict opioid dose requirements and strategies that improve the use of opioid dose titration in ED patients are needed.

Acknowledgments

The authors wish to thank the patients and ED providers who participated in this study; the ED enrollers and especially Sheri Friedberg for their efforts collecting the data; Maryann Countryman, Jessica Wilkie, and Sohug Mookerjee for their work on the study; and Dr Tim Quill for his thoughtful review of the manuscript.

This research was supported by The Mayday Fund. The authors report no potential conflicts of interest related to the study topic.

References

  1. Wilson JE, Pendleton JM. Oligoanalgesia in the emergency depart- ment. Am J Emerg Med 1989;7(6):620-3.
  2. Rupp T, Delaney KA. Inadequate analgesia in emergency medicine. Ann Emerg Med 2004;43(4):494-503.
  3. Todd KH, Ducharme J, Choiniere M, et al. Pain in the emergency department: results of the pain and emergency medicine initiative (PEMI) multicenter study. J Pain 2007;8(6):460-6.
  4. Pletcher MJ, Kertesz SG, Kohn MA, et al. Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA 2008;299:70-8.
  5. Stalnikowicz R, Mahamid R, Kaspi S, et al. Undertreatment of acute pain in the emergency department: a challenge. Int J Qual Health Care 2005;17:173-6.
  6. Bijur PE, Berard A, Esses D, et al. Lack of influence of patient self- report of pain intensity on administration of opioids for suspected long- bone fractures. J Pain 2006;7:438-44.
  7. Brown JC, Klein EJ, Lewis CW, et al. Emergency department analgesia for fracture pain. Ann Emerg Med 2003;42:197-205.
  8. Ritsema TS, Kelen GD, Pronovost PJ, et al. The national trend in quality of emergency department pain management for long bone fractures. Acad Emerg Med 2007;14:163-9.
  9. Rogovik AL, Rostami M, Hussain S, et al. Physician pain reminder as an intervention to enhance analgesia for extremity and clavicle injuries in pediatric emergency. J Pain 2007;8:26-32.
  10. American Pain Society. Principles of analgesic use in the treatment of acute pain and cancer pain. 5th ed. Glenview (Ill): APS8 2003. p. 15-8.
  11. O’Connor AB, Zwemer FL, Hays DP, et al. Outcomes after intravenous opioids in emergency patients: a prospective cohort analysis. Acad Emerg Med 2009;16:477-87.
  12. Bijur PE, Kenny MK, Gallagher EJ. Intravenous morphine at 0.1 mg/ kg is not effective for controlling severe pain in the majority of patients. Ann Emerg Med 2005;46:363-7.
  13. Lvovschi V, Aubrun F, Bonnet P, et al. Intravenous morphine titration to treat severe pain in the ED. Am J Emerg Med 2008;26 (6):676-82.
  14. Aubrun F, Monsel S, Langeron O, et al. Postoperative titration of intra- venous morphine in the elderly patient. Anesthesiology 2002;96:17-23.
  15. Cobaugh DJ, Spillane LL, Schneider SM. Research subject enroller program: a key to successful emergency medicine research. Acad Emerg Med 1997;4:231-3.
  16. Inturrisi CE. Clinical pharmacology of opioids for pain. Clin J Pain 2002;18:S3-S13.
  17. Gustein HB, Akil H. Opioid analgesics. In: Brunton L, Lazo J, Parker K, editors. Goodman and Gillman’s the pharmacological basis of therapeutics, 11th ed.New York (NY): McGraw-Hill; 2006. p. 547-90.
  18. Farrar JT, Young JP, LaMoreaux L, et al. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain 2001;94:149-58.
  19. Cepeda MS, Africano JM, Polo R, et al. What decline in pain intensity is Clinically meaningful to patients with acute pain? Pain 2003;105: 151-7.
  20. Dworkin RH, Turk DC, Wyrwich KW, et al. Interpreting the clinical importance of Treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. J Pain 2008;9:105-21.
  21. Stahmer SA, Shofer FS, Marino A, et al. Do quantitative changes in pain intensity correlate with pain relief and satisfaction? Acad Emerg Med 1998;5(9):851-7.
  22. Calvin A, Becker H, Biering P, et al. Measuring patient opinion of pain management. J Pain Symptom Manage 1999;18(1):17-26.
  23. Todd KH, Funk KG, Funk JP, et al. Clinical significance of reported changes in pain severity. Ann Emerg Med 1996;27(4):485-9.
  24. Tamayo-Sarver JH, Dawson NV, Cydulka RK, et al. Variability in emergency physician decisionmaking about prescribing opioid analgesics. Ann Emerg Med 2004;43:483-93.
  25. Miner J, Biros MH, Trainor A, et al. Patient and physician perceptions as risk factors for oligoanalgesia: a prospective observational study of the relief in the emergency department. Acad Emerg Med 2006;13:140-6.
  26. Chang AK, Bjur PE, Campbell CM, et al. Safety and efficacy of using 1 mg doses of intravenous hydromorphone in emergency department patients with acute severe pain: the “1 + 1” protocol. Ann Emerg Med 2009;54(2):221-5.
  27. Chang AK, Bijur PE, Meyer RH, et al. Safety and efficacy of hydromorphone as an analgesic alternative to morphine in acute pain: a randomized clinical trial. Ann Emerg Med 2006;48:164-72.
  28. Birnbaum A, Esses D, Bijur PE, et al. Randomized double-blind placebo-controlled trial of two intravenous morphine dosages (0.10 mg/kg and 0.15 mg/kg) in emergency department patients with moderate to severe acute pain. Ann Emerg Med 2007;49:445-53.
  29. Fry M, Holdgate A. Nurse-initiated intravenous morphine in the emergency department: efficacy, rate of adverse events and impact on time to analgesia. Emerg Med 2002;14:249-54.
  30. Ricard-Hibon A, Belpomme V, Chollet C, et al. Compliance with a morphine protocol and effect on pain relief in out-of-hospital patients. J Emerg Med 2008;34:305-10.
  31. Bounes V, Charpentier S, Houze-Cerfon CH, et al. Is there an ideal morphine dose for prehospital treatment of severe acute pain? A randomized, double-blind comparison of 2 doses. Am J Emerg Med 2008;26:148-54.
  32. Gordon DB, Dahl JL, Miaskowski C, et al. American Pain Society recommendations for improving the quality of acute and cancer pain management. Arch Intern Med 2005;165:1574-80.
  33. Marret E, Jurdi O, Zuffrey P, et al. Effects of nonsteroidal anti- inflammatory drugs on patient-controlled analgesia morphine side effects: meta-analysis of randomized controlled trials. Anesthesiology 2005;102:1249-60.
  34. Safdar B, Legutis LC, Landry K, et al. Intravenous morphine plus ketorolac is superior to either drug alone for treatment of acute renal colic. Ann Emerg Med 2006;48:173-81.
  35. Soliman HM, Melot C, Vincent JL. Sedative and analgesic practice in the intensive care unit: the results of a European survey. Br J Anaesth 2001;87:186-92.
  36. Bijur P, Berard A, Esses D, et al. Race, ethnicity, and management of pain from Long-bone fractures: a prospective study of two academic urban emergency departments. Acad Emerg Med 2008;15: 589-97.
  37. Quigley C, Wiffen P. A systematic review of hydromorphone in acute and chronic pain. J Pain Symptom Manage 2003;25: 169-78.

Appendix

Appendix Table 1 Comparison of the demographic, clinical, and treatment characteristics and the outcomes by opioid dose after exclusion of patients taking home opioids (the no home opioids cohort)

Morphine 2 mg n = 99

Morphine 4 mg

n = 223

Hydromorphone 1 mg n = 71

Hydromorphone 2 mg n = 29

Demographics

Age, y (median [IQR]) Female (n [%])

Race

White, non-Hispanic (n [%]) African American (n [%]) Hispanic (n [%])

Clinical characteristics

Initial pain score, 0-10 scale (median [IQR])

History of opioid allergy (n [%]) History of substance abuse (n [%]) History of chronic medical disease (n [%])

Pain diagnosis Abdominal pain (n [%]) Trauma (n [%])

Acute fracture (n [%]) Back pain (n [%]) Kidney stone (n [%])

Provider concern for patient stability (n [%])

Provider concern for drug-seeking (n [%])

Treatment characteristics

Initial equianalgesic dose, Mmg Weight-adjusted initial equianalgesic dose, Mmg/kg (median [IQR])

Total weight-adjusted equianalgesic dose, Mmg/kg (median [IQR]) Coadministered antiemetic (n [%]) Additional analgesic administered before postanalgesic survey (n [%]) Provider goal of full pain relief (n [%]) Analgesic outcomes

Postanalgesic pain score,

0-10 scale (median [IQR]) Change in pain score,

0-10 scale (median [IQR])

>=50% Pain score reduction (n [%])

>=30% Pain score reduction (n [%]) Desire for additional analgesics (n [%]) Pain relief, 1-5 scale (median [IQR]) Satisfaction with pain treatment,

1-6 scale (median [IQR]) Side effects

Serious side effects (n [%]) Any side effects (n [%])

Central nervous system side effects (n [%]) Sedation (n [%])

Dizziness (n [%])

39 (25-57)

39 (27-52)

40 (31-52)

38 (33-44)

64 (65)

133 (60)

44 (62)

12 (41)

62 (64)

150 (68)

47 (66)

21 (78)

27 (28)

55 (25)

16 (23)

6 (22)

5 (5)

13 (6)

6 (8)

0

8.0 (7.0-10)

8.0 (7.0-10)

10 (8.0-10)

10 (9.0-10)

9 (9)

31 (14)

14 (20)

10 (34) +

10 (10)

19 (9)

7 (10)

3 (10)

22 (22)

37 (17)

9 (13)

8 (28)

50 (51)

103 (46)

30 (42)

10 (34)

11 (11)

33 (15)

8 (11)

4 (14)

8 (8)

19 (9)

2 (3)

3 (10)

5 (5)

11 (5)

5 (7)

6 (21) ?

2 (2)

9 (4)

11 (15)

2 (7) +

31 (31)

71 (32)

21 (30)

10 (34)

18 (18)

23 (10)

13 (18)

12 (41) ?

2.0

4.0

6.7

13?

0.026 (0.022-

0.049 (0.043-

0.083 (0.069-0.098)

0.14 (0.13-0.17) ?

0.031)

0.061)

0.028 (0.023-

0.055 (0.045-

0.092 (0.073-0.12)

0.16 (0.13-0.21) ?

0.033)

0.068)

39 (39)

104 (47)

41 (58)

20 (69) ?

18 (18)

43 (19)

20 (28)

10 (34)

50 (51)

117 (52)

40 (56)

16 (55)

5.0 (2.0-8.0)

4.0 (2.0-7.0)

4.0 (2.0-7.0)

5.8 (2.5-8.0)

3.0 (1.0-5.0)

3.0 (1.0-5.0)

4.0 (2.0-7.0)

3.0 (1.5-5.5)

42 (42)

106 (48)

39 (55)

11 (38)

54 (55)

136 (61)

50 (70)

17 (59)

36 (36)

93 (42)

36 (51)

19 (66) ?

3.0 (3.0-4.0)

3.0 (3.0-4.0)

3.0 (3.0-5.0)

3.0 (2.0-4.0)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (3.5-6.0)

1 (1)

2 (1)

0

0

14 (14)

54 (24)

22 (31)

8 (28) ?

4 (4)

31 (14)

12 (17)

3 (10) ?

2 (2)

20 (9)

10 (14)

3 (10) ?

3 (3)

9 (4)

0

0

Appendix Table 2 Comparison of the demographic, clinical, and treatment characteristics and the outcomes by quartiles of weight-adjusted equianalgesic dose

Appendix Table 1 (continued)

Confusion (n [%]) Nausea (n [%]) Pruritis (n [%]) Dry mouth (n [%]) Headache (n [%])

Hypotension (n [%]) Disposition

Admitted (n [%])

Length of stay, d (median [IQR])

* P <= .05.

+ P b .01.

? P b .001.

Morphine 2 mg n = 99

0

4 (4)

0

0

0

1 (1)

Morphine 4 mg

n = 223

6 (3)

13 (6)

1 (0)

3 (1)

2 (1)

2 (1)

Hydromorphone 1 mg Hydromorphone 2 mg

n = 71

2 (3)

8 (11)

2 (3)

4 (6)

0

0

n = 29

0

1 (3)

2 (7) ?

1 (3) ?

1 (3)

0

28 (28)

0.50 (0.30-2.0)

57 (26)

0.43 (0.25-1.8)

17 (24)

0.40 (0.20-2.0)

6 (21)

0.40 (0.30-1.2)

Quartile 1

n = 171

Quartile 2

n = 171

Quartile 3

n = 177

Quartile 4

n = 172

Demographics

Age, y (median [IQR]) Female (n [%])

Race

White, non-Hispanic (n [%]) African American (n [%]) Hispanic (n [%])

Clinical characteristics

Initial pain score, 0-10 scale (median [IQR]) Any home opioid use (n [%])

Long-acting opioid use (n [%]) History of opioid allergy (n [%]) History of substance abuse (n [%])

History of chronic medical disease (n [%]) Pain diagnosis

Abdominal pain (n [%]) Trauma (n [%])

Acute fracture (n [%]) Back pain (n [%]) Kidney stone (n [%])

Provider concern for patient stability (n [%]) Provider concern for drug-seeking (n [%]) Treatment characteristics

Initial equianalgesic dose, Mmg (median [IQR])

Weight-adjusted initial equianalgesic dose, Mmg/kg (median [IQR])

Total weight-adjusted equianalgesic dose, Mmg/kg (median [IQR]) Hydromorphone (n [%])

Coadministered antiemetic (n [%]) Additional analgesic administered before postanalgesic survey (n [%]) Provider goal of full pain relief (n [%]) Analgesic outcomes

Postanalgesic pain score,

42 (28-54)

43 (30-55)

40 (28-50)

39 (30-49) ?

105 (61)

81 (47)

115 (65)

114 (66) +

111 (66)

119 (71)

117 (66)

103 (62)

44 (26)

36 (21)

46 (26)

47 (28)

8 (5)

8 (5)

9 (5)

13 (8)

8.0 (7.0-10)

9.0 (7.0-10)

9 (8.0-10)

10 (8.0-10)

43 (25)

49 (29)

52 (29)

76 (44) ?

4 (2)

5 (3)

11 (6)

25 (15) ?

21 (12)

22 (13)

38 (22)

41 (24) +

14 (8)

17 (10)

17 (10)

19 (11)

42 (25)

39 (23)

36 (20)

45 (26)

81 (47)

68 (40)

82 (46)

66 (38)

18 (11)

26 (15)

19 (11)

18 (10)

12 (7)

15 (9)

8 (5)

8 (5)

10 (6)

11 (6)

16 (9)

24 (14) ?

9 (5)

7 (4)

8 (5)

23 (13) +

42 (25)

50 (29)

56 (32)

51 (30)

29 (17)

21 (12)

36 (20)

53 (31) ?

2.0 (2.0-3.3)

4.0 (4.0-4.0)

4.0 (4.0-6.0)

6.7 (6.7-13.3) ?

0.028 (0.023-0.033)

0.046 (0.042-0.049)

0.064 (0.060-0.070)

0.12 (0.092-0.16) ?

0.030 (0.025-0.035)

0.049 (0.044-0.053)

0.065 (0.061-0.075)

0.13 (0.098-0.18) ?

7 (4)

18 (11)

42 (24)

137 (80) ?

67 (39)

78 (46)

80 (45)

94 (55) ?

35 (20)

47 (27)

40 (23)

50 (29)

85 (50)

97 (57)

96 (54)

88 (51)

5.0 (2.0-7.0)

5.0 (3.0-8.0)

5.0 (3.0-7.0)

5.0 (3.0-8.0)

Appendix Table 3 Comparison of the demographic, clinical, and treatment characteristics and the outcomes by quartiles of weight-adjusted equianalgesic dose after exclusion of patients taking home opioids (the no home opioids cohort)

Appendix Table 2 (continued)

Quartile 1

n = 171

Quartile 2

n = 171

Quartile 3

n = 177

Quartile 4

n = 172

0-10 scale (median [IQR]) Change in pain score,

0-10 scale (median [IQR])

>=50% Pain score reduction (n [%])

>=30% Pain score reduction (n [%]) Desire for additional analgesics (n [%]) Pain relief, 1-5 scale (median [IQR]) Satisfaction with pain treatment,

1-6 scale (median [IQR]) Side effects

Serious side effects (n [%]) Any side effects (n [%])

Central nervous system side effects (n [%]) Sedation (n [%])

Dizziness (n [%]) Confusion (n [%]) Nausea (n [%]) Pruritis (n [%]) Dry mouth (n [%]) Headache (n [%])

Hypotension (n [%]) Disposition

Admitted (n [%])

Length of stay, d (median [IQR])

3.0 (1.75-5.0)

2.0 (1.0-5.0)

3.0 (1.0-5.0)

3.5 (1.0-6.0)

80 (47)

65 (38)

75 (42)

75 (44)

103 (60)

93 (54)

102 (58)

104 (60)

65 (38)

92 (48)

89 (50)

96 (56) +

3.0 (2.0-4.0)

3.0 (3.0-4.0)

3.0 (2.0-4.0)

3.0 (2.0-4.0)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (3.5-6.0)

1 (1)

1 (1)

1 (1)

1 (1)

31 (18)

38 (22)

43 (24)

47 (27)

12 (7)

22 (13)

25 (14)

24 (14)

7 (4)

15 (9)

18 (10)

20 (12)

8 (5)

5 (3)

3 (2)

4 (2)

1 (1)

5 (3)

6 (3)

4 (2)

8 (5)

8 (5)

10 (6)

9 (5)

1 (1)

1 (1)

4 (2)

5 (3)

1 (1)

4 (2)

2 (1)

4 (2)

2 (2)

1 (1)

2 (1)

1 (1)

1 (1)

1 (1)

1 (1)

1 (1)

40 (23)

49 (29)

48 (27)

55 (32)

0.43 (0.30-1.5)

0.50 (0.30-2.4)

0.40 (0.20-2.0)

0.69 (0.30-2.2)

* P <= .05.

+ P b .01.

? P b .001.

Quartile 1 n = 116

Quartile 2

n = 123

Quartile 3 n = 114

Quartile 4 n = 118

Demographics

Age, y (median [IQR])

40.5 (26.5-53)

39 (27-53)

39 (28-50)

39.5 (30-51)

Female (n [%])

68 (59)

60 (49)

77 (68)

71 (60) ?

Race

White, non-Hispanic (n [%])

74 (64)

86 (71)

70 (62)

86 (74)

African American (n [%])

32 (28)

26 (21)

33 (29)

22 (19)

Hispanic (n [%])

6 (5)

7 (6)

8 (7)

6 (5)

Clinical characteristics

Initial pain score, 0-10 scale (median [IQR])

8.0 (7.0-10)

9.0 (7.0-10)

8.0 (7.0-10)

9.5 (8.0-10) +

History of opioid allergy (n [%])

10 (9)

14 (11)

20 (18)

28 (24) +

History of substance abuse (n [%])

11 (9)

12 (10)

9 (8)

16 (14)

History of chronic medical disease (n [%])

24 (21)

20 (16)

17 (15)

18 (15)

Pain diagnosis

Abdominal pain (n [%])

59 (51)

48 (39)

56 (49)

47 (40)

Trauma (n [%])

12 (10)

23 (19)

11 (10)

23 (19) ?

Acute fracture (n [%])

8 (7)

12 (10)

7 (6)

8 (7)

Back pain (n [%])

6 (5)

6 (5)

6 (5)

10 (8)

Kidney stone (n [%])

4 (3)

6 (5)

6 (5)

15 (13) ?

Provider concern for patient stability (n [%])

32 (28)

39 (32)

36 (32)

39 (33)

Provider concern for drug-seeking (n [%])

20 (17)

14 (12)

14 (12)

25 (21)

Appendix Table 3 (continued)

Quartile 1 n = 116

Quartile 2

n = 123

Quartile 3 n = 114

Quartile 4 n = 118

Treatment characteristics

Initial equianalgesic dose,

2.0 (2.0-2.0)

4.0 (4.0-4.0)

4.0 (4.0-6.0)

6.7 (6.7-13.3) ?

Mmg (median [IQR])

Weight-adjusted initial

0.027 (0.022-0.032)

0.044 (0.041-0.049)

0.061 (0.057-0.067)

0.097 (0.081-0.13) ?

equianalgesic dose,

Mmg/kg (median [IQR])

Total weight-adjusted

0.029 (0.024-0.033)

0.046 (0.042-0.050)

0.062 (0.058-0.068)

0.11 (0.084-0.16) ?

equianalgesic dose,

Mmg/kg (median [IQR])

Hydromorphone (n [%])

1 (1)

7 (6)

18 (16)

81 (69) ?

Coadministered antiemetic (n [%])

49 (42)

58 (47)

51 (45)

74 (63) +

Additional analgesic administered

23 (20)

32 (26)

21 (18)

32 (27)

before postanalgesic survey (n [%])

Provider goal of full pain relief (n [%])

55 (47)

70 (57)

60 (53)

62 (53)

Analgesic outcomes

Postanalgesic pain score,

5.0 (2.0-7.0)

5.0 (2.0-8.0)

5.0 (2.0-7.0)

5.0 (2.0-8.0)

0-10 scale (median [IQR])

Change in pain score,

3.0 (1.5-5.0)

3.0 (1.0-5.0)

3.0 (2.0-5.0)

4.0 (1.0-6.0)

0-10 scale (median [IQR])

>=50% Pain score reduction (n [%])

55 (47)

54 (44)

53 (46)

59 (50)

>=30% Pain score reduction (n [%])

68 (59)

70 (57)

72 (63)

74 (63)

Desire for additional analgesics (n [%])

44 (38)

52 (42)

51 (45)

63 (53)

Pain relief, 1-5 scale (median [IQR])

3.0 (3.0-4.0)

3.0 (3.0-4.0)

3.0 (3.0-4.5)

3.0 (3.0-4.0)

Satisfaction with pain treatment,

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

5.0 (4.0-6.0)

1-6 scale (median [IQR])

Side effects

Serious side effects (n [%])

1 (1)

1 (1)

0

1 (1)

Any side effects (n [%])

21 (18)

31 (25)

26 (23)

37 (31)

Central nervous

7 (6)

17 (14)

13 (11)

19 (16)

system side effects (n [%])

Sedation (n [%])

2 (2)

12 (10)

9 (8)

15 (13) +

Dizziness (n [%])

5 (4)

5 (4)

2 (2)

1 (1)

Confusion (n [%])

1 (1)

3 (2)

3 (3)

3 (3)

Nausea (n [%])

6 (5)

6 (5)

7 (6)

7 (6)

Pruritis (n [%])

0

1 (1)

2 (2)

5 (4)

Dry mouth (n [%])

1 (1)

3 (2)

1 (1)

4 (3)

Headache (n [%])

1 (1)

1 (1)

2 (2)

1 (1)

Hypotension (n [%])

1 (1)

1 (1)

0

1 (1)

Disposition

Admitted (n [%])

29 (25)

33 (27)

24 (21)

35 (30)

Length of stay, d (median [IQR])

0.43 (0.30-2.0)

0.40 (0.22-2.0)

0.40 (0.20-1.7)

0.60 (0.30-2.0)

* P <= .05.

+ P b .01.

? P b .001.

adjusted

opioid dose ?

>=50% Pain score reduction

Unadjusted Adjusted odds ratio odds ratio

Any side effects

Unadjusted Adjusted

odds ratio odds ratio

>=50% Pain

Unadjusted Adjusted odds ratio odds ratio

Any side effects

Unadjusted Adjusted

odds ratio odds ratio

(95% CI)

(95% CI)

(95% CI) (95% CI)

(95% CI)

(95% CI)

(95% CI) (95% CI)

Quartile 1

Reference

Reference

Reference Reference

Reference

Reference

Reference Reference

Quartile 2

0.70

0.75

1.29 1.27

0.87

0.86

1.52 1.48

(0.45-1.07)

(0.48-1.18)

(0.76-2.19) (0.74-2.16)

(0.52-1.44)

(0.50-1.46)

(0.82-2.84) (0.79-2.78)

Quartile 3

0.84

0.86

1.45 1.48

0.96

0.86

1.34 1.33

(0.55-1.28)

(0.55-1.34)

(0.86-2.44) (0.87-2.49)

(0.57-1.62)

(0.50-1.48)

(0.70-2.55) (0.70-2.54)

Quartile 4

0.88

1.12

1.70 1.78

1.11

1.20

2.07 2.13

(0.57-1.35)

(0.70-1.79)

(1.02-2.84) + (1.05-3.00)

(0.66-1.85)

(0.70-2.08)

(1.12-3.81) + (1.13-4.01) +

Appendix Table 4 Unadjusted and adjusted odds ratios of categorical outcomes associated with the weight-adjusted equianalgesic dose quartiles

Weight- All patients No home opioids

score reduction

‘All patients’ indicates the entire cohort; ‘no home opioids’, the cohort remaining after excluding the patients taking home opioids.

* The median weight-adjusted dosages for the all patients cohort were 0.028 Mmg/kg for quartile 1, 0.046 Mmg/kg for quartile 2, 0.064 Mmg/kg for

quartile 3, and 0.12 Mmg/kg for quartile 4. The median weight-adjusted dosages for the no home opioids cohort were 0.027 Mmg/kg for quartile 1, 0.044 Mmg/kg for quartile 2, 0.061 Mmg/kg for quartile 3, and 0.097 Mmg/kg for quartile 4.

+ P <= .05 for comparison with quartile 1.

Appendix Fig. 1 Distribution of the pain score changes resulting from the most commonly administered opioid dosages. All patients indicates the entire cohort; no home opioids, the cohort remaining after excluding the patients taking home opioids.

Appendix Fig. 2 Relationship between weight-adjusted equi- analgesic dosage and pain score change. Note that 3 out-lying values (N0.5 Mmg/kg) are not included to better show the remaining data. All patients indicates the entire cohort; no home opioids, the cohort remaining after excluding the patients taking home opioids.

Appendix Fig. 3 Box plots demonstrating the pain score improvement by quartile of weight-adjusted opioid dosage. The boxes show the interquartile range around the median line. The surrounding bars demonstrate the 10th and 90th percentile values. All patients indicates the entire cohort; no home opioids, the cohort remaining after excluding the patients taking home opioids. The restricted subset consists of patients meeting each of the following criteria: no long-acting opioid use, not considered potentially drug-seeking, no additional analgesics, and age of less than 65 years. The most restricted subset is the restricted subset patients who had an initial pain score of 10 (the mode). P N .05 for all within-subset comparisons except P = .01 for the comparisons in the restricted subset and P = .02 for comparisons in the most restricted subset.

Leave a Reply

Your email address will not be published. Required fields are marked *