Effect of Combining Naloxone and Morphine for Intravenous Patient-controlled Analgesia

The study was approved by the hospital's institutional review board (Cairns Base Hospital, Cairns, Queensland, Australia), and written informed consent was obtained from all participants. Female patients aged older than 18 yr who were scheduled to undergo abdominal or vaginal hysterectomy for benign disease under general anaesthesia were eligible for inclusion. Patients with known morphine allergy or chronic opioid use were excluded. Patients were instructed before surgery on the use of PCA and the nature of symptom evaluation.

A standard general anesthetic was given, comprising 1–3 mg midazolam, 1–2 μg/kg fentanyl, 1–3 mg/kg propofol, 0.1–0.2 mg/kg morphine, nitrous oxide, and sevoflurane or isoflurane. All patients received vecuronium or rocuronium for tracheal intubation and reversal with neostigmine and atropine. No prophylactic antiemetics or nonsteroidal antiinflammatory drugs were permitted. In the postanesthesia care unit, analgesia was provided with morphine in 1- to 2-mg increments until the patient was comfortable and sufficiently alert to use PCA.

Patients were stratified by operation (abdominal or vaginal hysterectomy) and randomized to naloxone or placebo (saline) groups according to a system of random numbers and a sealed envelope. The study was double blind, with the patient, anesthesiologist, nursing staff, and observers all unaware of the randomization. Two nurses, otherwise uninvolved in the patient's care, opened the envelope and added the appropriate solution to the PCA syringe. Preservative-free morphine, 60 mg, was added to saline, with or without 0.8 mg naloxone, to a total of 30 ml. The naloxone dose was based on the amounts infused over a day by Gan et al. , 2and the predicted 24-h morphine use was based on information from our Acute Pain Service database. Parameters for PCA were a 1-mg bolus of morphine with a 5-min lockout and no background infusion. On the surgical ward, 1 g paracetamol (acetaminophen) every 4 hours was permitted for inadequate analgesia.

In the event of nausea or vomiting, antiemetic treatment was offered and was provided sequentially as 10 mg intravenous metoclopramide, 4 mg ondansetron, and 1 mg droperidol until symptoms were controlled. Troublesome pruritus was treated with 10 mg intravenous promethazine.

At 6 and 24 h postoperatively, patient recall of the severity of pain since the operation, at rest and with movement, was assessed by a 0- to 100-mm visual analog score and a verbal rating score of none, mild, moderate, or severe. Nausea and itch were assessed in the same way by verbal rating score, and the patients were asked if they had vomited. The level of sedation was assessed at these times by the researcher rather than by patient report, using the same categorical scale. Moderate sedation was defined as very sleepy but rousable; severe sedation was defined as difficult to rouse or unrousable. The amount of morphine used and the requirements for supplemental analgesia or symptomatic treatment over 24 h were also recorded.

For determination of sample size, we considered a reduction in the incidence of nausea the most relevant clinical outcome. Gan et al.  2showed a decrease in nausea from 80% of patients with placebo to 35% and 45% with the two treatment arms of their study. We estimated a slightly more conservative decrease, from 75% nausea with placebo to 45% with treatment. Forty-six patients in each group would give an 80% chance of detecting this effect with a P  value of 0.05. In the event of patient withdrawal, further patients were to be recruited until 92 patients had completed the study. Data were stored on Access 2000 (Microsoft Corporation, Redmond, WA) and analyzed with SigmaStat 2.0 (SPSS Inc., Chicago, IL). Interval and continuous data were assessed for normality using the Kolmogorov-Smirnov test, with Student t  test and Mann–Whitney U test applied for normal and skewed data, respectively. For categorical data, the chi-square (without the Yates correction) or Fisher exact test was used as appropriate.

Ninety-six patients were randomized, and 92 completed the study. One from each group required early reoperation for bleeding, a patient from the naloxone group experienced a suspected antibiotic allergy during surgery, and a patient from the control group underwent a lesser operation. All patients were classified as having American Society of Anesthesiologists physical status I or II. There were no baseline differences between the treatment and control groups (table 1).

There were no differences between the groups in the incidence of nausea, itching, sedation, or the need for symptomatic treatment at either assessment time. No patients were severely sedated (tables 2 and 3). In addition, there were no differences in these outcomes between the two operation groups (abdominal or vaginal hysterectomy). The incidence of nausea in patients receiving more than the median amount of morphine over 24 h was exactly the same as those receiving less.

There were no differences between the groups in terms of pain scores (both visual analog score and verbal rating score), amount of morphine used, or need for supplemental analgesia at either assessment time (table 4).

Among patients receiving naloxone, the median amount received over 24 h was 0.59 mg, which equated to 0.38 μg · kg⁻¹· h⁻¹over 24 h. The median amount of naloxone received over the first 6 h equated to 0.56 μg · kg⁻¹· h⁻¹and for the last 18 h equated to 0.32 μg · kg⁻¹· h⁻¹.

In three previous studies, the use of opioid antagonists with intravenous PCA morphine resulted in either beneficial or adverse effects. 2,3,8In contrast, we found no treatment effect from adding naloxone to PCA morphine.

Gan et al.  2used fixed-rate naloxone infusions at 0.25 μg · kg⁻¹· h⁻¹(low rate) and 1.0 μg · kg⁻¹· h⁻¹(high rate) in patients undergoing abdominal hysterectomy. They reported a decrease in nausea from 80% with placebo to 45% (low rate) and 35% (high rate) and a decrease in pruritus from 55% with placebo to 25% (low rate) and 20% (high rate). They also reported a decrease in opioid requirements compared to placebo with the low-rate infusion but no difference from placebo with the high-rate infusion.

The reason for the disparity in outcomes between our study and that of Gan et al.  2is unclear, though the differences in study design raise several possibilities. Conceivably, differences in symptom assessment between our study and that of Gan et al.  2may have influenced the results. Gan et al.  2used multiple contemporaneous measures, whereas we used patient recall of symptoms at 6 and 24 h. Although the two modes of assessment are not interchangeable, patient recall of nausea is generally reliable, and the overall incidence of symptoms is similar with the two methods. 9,10Indeed, we found an incidence of nausea and pruritus very similar to the placebo group in the study of Gan et al.  2We also found no difference in the requirement for rescue medications between groups, in contrast to the findings of Gan et al.  2 

The total amount of morphine and the median total dose of naloxone received by our patients are similar to the amounts received by the patients studied by Gan et al.  2However, because we administered naloxone mixed with PCA morphine, in contrast to the separate infusion technique of Gan et al. , 2factors other than total drug dose may be influential.

We believe there is no pharmaceutical incompatibility when morphine and naloxone are combined in saline, which might have offered an explanation for our negative findings. Although we are unaware of direct information on naloxone stability in this situation, in two published studies, naloxone was shown to have clinical activity when similarly combined with morphine. 5,8 

Using methods similar to ours, Cepeda et al.  8also combined naloxone with PCA morphine in male and female patients undergoing a range of surgical treatments. They used a lower bolus dose of naloxone of 6 μg for each 1 mg morphine, compared to 13.3 μg per 1 mg morphine in our study, and their total 24-h naloxone dose, equating to 0.10 μg · kg⁻¹· h⁻¹, was much lower than the 0.38 μg · kg⁻¹· h⁻¹in our study. 8Despite these tiny doses, they found an increase in morphine requirements and more analgesic failures in the naloxone group. 8They found no difference in opioid-related side effects, though admittedly in patients at lower risk of such complications than those undergoing hysterectomy. 8 

The lack of benefit from naloxone in our study and that of Cepeda et al.  8may reflect the intermittent administration of naloxone in these studies, in contrast to the continuous infusion of Gan et al.  2Naloxone has a half-life of 55 min, so administration by PCA would have resulted in fluctuation of naloxone concentrations. 11Perhaps a constant concentration of naloxone is required for it to have a beneficial effect.

Joshi et al.  3,11studied the longer-acting opioid antagonist nalmefene (half-life, 108 min) in patients receiving intravenous PCA morphine. After a single bolus of 15 or 25 μg, they found a decrease in patient recall of opioid side effects and pain severity, as well as a decreased requirement for symptomatic treatment. 3However, these promising results have not been repeated or confirmed in other situations. 12 

In summary, we found no benefit from adding naloxone to intravenous PCA morphine in women undergoing hysterectomy. We believe that in view of the limited and contradictory data available, further research is necessary before the routine clinical use of opioid antagonists with intravenous opioids can be recommended. If any benefit exists from this strategy, it would appear to be more likely with naloxone infusions or long-acting opioid antagonists.