Incidence of constipation associated with long-acting opioid therapy: a comparative study.

Background: Opioid therapy plays a key role in the management of chronic pain. Constipation is one of the more frequently occurring adverse effects associated with opioid therapy.

Methods: A retrospective cohort design study was conducted to determine the incidence of constipation in chronic pain patients who received three different long-acting opioids (transdermal fentanyl, oxycodone HCI controlled-release [CR], or morphine CR) for malignant or nonmalignant chronic pain. The data source was claims data (January 1996 through March 2001) from a 20% random sample of the California Medicaid (Medi-Cal) database. Claims data were from adult patients with chronic pain (malignant or nonmalignant) who had no prior diagnosis of constipation and no prior usage of long-acting opioids for at least 3 months before the observation period. Patients were followed for at least 3 months after the initiation of opioid therapy. ICD-9 code for diagnosis of constipation was the main outcome variable. Crude rates of constipation, annual incidence density, relative risk, and adjusted odds ratios were compared.

Results: A total of 1,836 patients (601 receiving transdermal fentanyl, 721 receiving oxycodone CR, and 514 receiving morphine CR) were included in the analysis. Crude (unadjusted) rates of constipation were 3.7% for transdermal fentanyl, 6.1% for oxycodone CR, and 5.1% for morphine CR (P > 0.05). Transdermal fentanyl had a lower annual incidence density and risk of constipation than oxycodone CR and morphine CR (P > 0.05). After adjusting for confounding variables, including race and supplemental opioid use, the adjusted risk of constipation was 78% greater in the oxycodone CR group (P = 0.0337) and 44% greater in the morphine CR group (P = 0.2242) than in the transdermal fentanyl group.

Conclusion: In this population, patients receiving transdermal fentanyl had a lower risk of developing constipation compared with those receiving oxycodone CR or morphine CR.

Key Words: constipation, fentanyl transdermal system, morphine controlled-release, opioid analgesics, oxycodone controlled-release

**********

Constipation is a symptom of several diseases. A variety of drugs--such as tricyclic antidepressants, aluminum-containing antacids, opioids, and several others--may be associated with constipation. In addition to its negative effect on quality of life, severe and persistent constipation may lead to serious medical sequelae, such as bowel obstruction or fecal impaction.

Opioids have been widely used to treat malignant pain, and their use to manage nonmalignant pain has also been increasing in recent years. Patients may develop a tolerance to opioid-related adverse effects, but constipation is often persistent. Depending upon the receptor affected (eg, mu, kappa, delta), opioid analgesics may alter normal peristaltic motion, affect water absorption, or slow intestinal transit. (1)

The objective of the present study was to compare constipation rates across three long-acting opioids: fentanyl transdermal system, oxycodone HCI controlled-release (CR), and morphine CR. Because these long-acting opioids have different pharmacokinetic profiles and different affinities with which they bind to receptors, one might speculate that the incidence of constipation will also vary among these drugs. We used the Medi-Cal system to examine patients with an International Statistical Classification of Diseases, 9th Revision (ICD-9) diagnosis of constipation, because this database contains a large number of patients with integrated (ie, pharmacy and medical) claims data.

Methods

Study Design and Patient Population

This was a retrospective cohort study designed to compare incidence rates of constipation in chronic pain patients receiving one of three long-acting opioids (transdermal fentanyl, oxycodone CR, and morphine CR) (Fig. 1). Analysis was based on claims data from Medi-Cal patients with chronic malignant and nonmalignant pain.

Claims data were obtained from a 20% random sample of prescriptions from the Medi-Cal database from January 1996 through the end of June 2001. Eligible patients were those adults ([greater than or equal to]18 yr of age) who were continuously enrolled in Medi-Cal for a period of at least 6 months, who had no primary ICD-9 claim(s) for constipation, and who had no prescription claim for transdermal fentanyl, oxycodone, or morphine during the previous 3 months. Excluded were those patients with claims for two or more of the opioids of interest, prescribed simultaneously, during the study period; concomitant use of opioids other than transdermal fentanyl, oxycodone CR, and morphine CR (eg, codeine), however, was not grounds for exclusion. From the pool of eligible patients, those who received a new prescription for transdermal fentanyl, oxycodone CR, or morphine CR (the "index" claim) during the study period and who had follow-up data for a minimum of 3 months were selected as participants and their claims data were analyzed. Length of the observation period was variable; for all patients, observation started with the date of the index claim and continued to the first date of any of the following: a constipation event, the end of continuous enrollment in Medi-Cal, switch to another long-acting opioid, or the end point of the database (ie, June 2001).

Data Analysis

Patients in the study cohort were divided into three independent groups, based on their index prescription for one of the long-acting opioids of interest: transdermal fentanyl, oxycodone CR, or morphine CR (the "index opioids"). The primary outcome variable was the presence or absence of an ICD-9 code for constipation during the observation period for each given patient. Crude incidence rates for constipation were determined within each treatment group by taking the number of patients with constipation events (within each treatment group) and dividing by the total number of patients (within each treatment group). In addition, to account for length of index opioid exposure, incidence density and relative risk ratios were computed. Within each treatment group, incidence density was set equal to the number of new cases of constipation divided by the total number of exposure days; this gave the rate of incidence per day of exposure. From this, an annual incidence density was calculated by multiplying by 365.25 (days). Relative risk ratios for oxycodone CR and morphine CR were computed by dividing their respective annual incidence density figures by the annual incidence density for the transdermal fentanyl group.

[FIGURE 1 OMITTED]

To account for the potential confounding influence on incidence of constipation by other variables besides index opioid use, a number of demographic and medical factors were included in the data analysis. These included age, sex, race, cancer diagnosis, surgery status, and depression. Three types of mean opioid doses per day ("morphine equivalents") were also included in the analysis: 1) dose per day of index opioid, 2) dose per day of supplemental opioid (any opioid analgesic other than one of the three index opioids), and 3) dose per day of combined opioids taken (sum of index opioid dose and supplemental opioid dose). These were categorized as low, medium, or high based on their respective frequency distribution. For mean dose per day of index opioids, low morphine equivalents were defined as less than 61 mg, medium equivalents were greater than 60 mg and less than 121 mg, and high equivalents were greater than 120 mg. For supplemental opioids, low morphine equivalents were defined as less than or equal to 12 mg, medium equivalents were between 12.1 and 58.7 mg, and high equivalents were equal to or greater than 58.8 mg. If patients received no supplemental drugs during the observation period, they were placed into a fourth category ("None"). For combined opioid dose per day morphine equivalents, the respective cutoffs were less than or equal to 74 mg for low, between 75 and 159.8 mg for medium, and equal to or greater than 159.9 mg for high. Crude incidence and annual incidence density for constipation were calculated and examined based on these variables as well.

Statistical Analysis

Crude incidence rates were compared among the three treatment groups using [chi square]. To statistically test for differences in annual incidence density rates, relative risk ratios with 95% confidence intervals (CI) were computed and compared. Adjusted odds ratios (OR) were computed based on comparison with the following reference groups: treatment group (transdermal fentanyl), age (18-44 yr), sex (female), race (white), cancer (absent), index morphine equivalents (low), supplemental morphine equivalents (none), and combined morphine equivalents (low).

Potential confounding variables were identified using a two-step approach. First, treatment group (eg, index opioid group) was crossed with all other study variables (excluding constipation) to identify possible differences ("imbalances") among the index opioid groups. Second, control variables were examined for possible relationships with constipation. Only those variables that were significant in both of these steps were incorporated into a logistic regression model, along with variables expressing treatment group and length of exposure ([alpha] level <0.05 [two-tailed]). Adjusted OR with 95% CIs as well as parameter estimates, standard errors, and P values were provided for all variables entered into the logistic regressions.

Results

Patient Characteristics

A total of 1,836 patients (transdermal fentanyl: n = 601; oxycodone CR: n = 721; morphine CR: n = 514) met the eligibility criteria and were included in the analysis. The average age of the overall patient population was 58.8 ([+ or -]17.1) years; 62% of the patients were female, and the majority of patients (65.9%) were white. The percentage of patients with cancer (n = 532) was 29%, and 42.3% had undergone surgery (n = 776). For the group as a whole, patients had a mean of 313.4 days of exposure to one of the long-acting index opioids. The demographic characteristics of the patient population by treatment group are shown in Table 1. There were statistically significant differences among the treatment groups according to age, gender, race, proportions of patients with cancer and depression, as well as opioid dose (ie, morphine equivalents) and exposure (P < 0.001). The transdermal fentanyl group had the highest mean age (66.3 yr), the lowest proportion of male patients (29.3%), the shortest length of index opioid exposure (276.6 d), and was prescribed the lowest combined opioid morphine equivalents. By contrast, significantly more patients in the oxycodone CR group had depression (15.3%) than either the transdermal fentanyl (5.8%) or morphine CR (11.7%) groups. Oxycodone patients also had significantly higher index and combined morphine equivalents than patients in either the transdermal fentanyl or morphine CR groups (Table 1).

Constipation Incidence and Unadjusted Risk

A total of 92 patients had at least one reported constipation event, yielding an overall crude rate of constipation of 5.0%. Although the crude rate of constipation was somewhat lower in the transdermal fentanyl group (3.7%), compared with the oxycodone CR (6.1%) and morphine CR (5.1%) groups (Fig. 2; Table 2), no statistically significant differences in crude constipation rates were observed among the treatment groups ([chi square] [2 df] = 4.11, P = 0.1281). Similarly, annual incidence density for constipation was slightly lower with transdermal fentanyl (4.8%) than with oxycodone CR (6.8%) and morphine CR (5.5%), but no statistically significant differences were observed among the groups (Table 2). These annual incidence densities indicate that relative risk of constipation was lower with transdermal fentanyl than with oxycodone CR (41% less) and morphine CR (13% less); given the observed 95% CIs, however, neither of these differences in relative risk was significant.

Relationship of Control Variables to Constipation Incidence

No significant relationships were observed between annual incidence density of constipation and a number of control variables examined, including patient age, gender, and presence or absence of cancer or depression. Neither was annual incidence density of constipation found to be related to low, medium, or high index morphine equivalents or combined morphine equivalents (ranging between 5.1% and 7.1% per year). Significant relationships, however, were observed between annual incidence density of constipation and race, occurrence of surgery, and use of supplemental opioids. A higher annual incidence density of constipation was observed among Hispanic patients and those marked "Missing/Other" for race (8.0% and 10.3%, respectively) than was observed for whites or blacks (4.9% and 3.9%, respectively). For the "Missing/Other" race group, the rate of constipation was significantly higher (relative to whites), with an OR of 2.11 (95% CI = 1.34, 3.31). Patients who had not undergone surgery had higher annual incidence density for constipation (7.4% per year) than patients who had undergone surgery (4.6% per year). Patients who had not used supplemental opioids as "rescue" medication had significantly higher annual incidence density of constipation (9.2% per year) than patients who used any dose of such supplemental therapy (3.5, 5.3, and 4.6% per year, respectively, for low, medium, and high supplemental opioid morphine equivalents). Analysis showed that only race and supplemental opioid use were related to both index treatment and annual constipation incidence density. To account for the influence of these two potentially confounding variables in estimates of constipation risk, they were consequently included in the final logistic regression model.

[FIGURE 2 OMITTED]

Adjusted Risk for Constipation. After controlling for differences in race, supplemental opioid use, and number of days of opioid exposure, use of oxycodone CR was identified as a significant independent predictor of constipation. Compared with patients who received transdermal fentanyl, those administered oxycodone CR exhibited a 78% increased risk of constipation (oxycodone CR adjusted OR = 1.78; 95% CI = 1.05, 3.03; P = 0.0337; Fig. 3). The risk of constipation was also increased with morphine CR (44%) compared with transdermal fentanyl, but this difference was not statistically significant (morphine CR adjusted OR = 1.44; 95% CI = 0.80, 2.60; P = 0.2242; Fig. 3).

Discussion

After adjusting for the potential confounding variables of race and supplemental opioid use, a significant increase (78%) was found in the risk of constipation in chronic pain patients treated with oxycodone CR relative to transdermal fentanyl, and a nonsignificant (44%) increased risk of constipation was found in patients treated with morphine CR. This finding is supported by the unadjusted outcome measures; a higher crude rate and annual incidence density of constipation among patients who received oxycodone CR or morphine CR compared with those who received transdermal fentanyl were observed. This latter trend, however, did not reach statistical significance. These data suggest that transdermal fentanyl may be associated with a lower risk of constipation than oxycodone CR and possibly other long-acting opioids, such as morphine CR.

Other studies have also demonstrated that, compared with sustained-release oral morphine, transdermal fentanyl is associated with a lower incidence of constipation as well as decreased laxative use. (2-7) Factors that may contribute to the reduced effect of transdermal fentanyl on intestinal activity include differences in local drug effects on the gut (eg, half-life in intestinal tissue and absorption into the intestinal lumen), first-pass metabolism, routes of administration, or steady-state plasma levels. (7-9) Furthermore, recent molecular studies have demonstrated the presence of multiple [mu]-receptor types, suggesting that opioids generate their analgesic, and possibly intestinal, effects through subtle differences in binding selectivities to different subtypes. (10) It seems likely that a combination of these factors contributes to the reduced effect of transdermal fentanyl on intestinal activity.

[FIGURE 3 OMITTED]

Our analysis offers a number of advantages compared with previous studies of long-acting opioids and their common adverse effects. Patients were followed for at least 3 months and up to 2 years, representing a course of long-term opioid therapy. In contrast, most clinical trials involving the long-acting opioids studied here, including those concerning management of chronic pain, are typically of much shorter duration, lasting for as little as 2 weeks. (7), (11), (12) We also calculated annual incidence density, which takes into account the length of exposure time to long-acting opioids, and OR adjusted for other significant predictors of constipation such as race and supplemental opioid use, which allows a more direct and valid comparison among agents. Moreover our observations are based on actual clinical use of long-acting opioids, without the constraints of clinical trial procedures. Drug choices, dosing, and duration of drug administration are likely to reflect those patterns typically seen in a diverse patient population being treated for chronic pain arising from a variety of causes, including cancer or surgery.

Certain limitations inherent with the use of a claims analysis approach are also likely to have influenced our observations and possibly caused them to err toward being conservative. For instance, the crude rate of constipation (5.0%), defined by the presence of a corresponding ICD-9 code, is probably an underestimate of actual incidence of constipation and probably represents cases of severe and/or unremitting constipation; less severe cases were most likely not detected by our analysis. Furthermore, our estimates of mean opioid dose per day may not reflect actual medication usage; on any given day, patients may have used more or less opioid than was prescribed. Because these data were derived from a Medicaid population, our findings may not necessarily be generalizable to other patient populations. Use of the Medi-Cal database, however, afforded the analysis of a large patient cohort that received long-acting opioids for commonly encountered pain management problems, such as pain arising from surgery and cancer. This study did not control for other medications in the preobservation or observation period that could contribute to constipation.

Conclusions

Our observations indicate that constipation continues to present a significant challenge in the use of long-acting opioids for chronic pain. Management guidelines state that constipation prevention rather than amelioration is the treatment goal for patients using opioids for pain management, and a number of clear steps toward this goal have been outlined. (13), (14) Our findings suggest that transdermal fentanyl provides an opioid analgesic treatment option that may reduce the risk of constipation relative to oxycodone CR and possibly other long-acting opioids.

Key Points

* Constipation is one of the more frequently occurring adverse effects associated with opioid therapy.

* The incidence of constipation in study participants varied depending on the type of opioid therapy administered.

Well done is better than well said.

--Benjamin Franklin

Table 1. Patient characteristics by treatment group (a)

                                   Treatment group

Characteristic (mean         Transdermal      Oxycodone HCI CR
[SD] or %)                fentanyl (n = 601)    (n = 721)

Age, yr                       66.3 (17.8) (b)    54.3 (15.7)

Gender, % male                29.3 (b)           40.1

Race, %
  White                       65.6               70.3
  Black                       11.3                6.1
  Hispanic                     3.8                2.9
  Missing/Other data          19.3               20.7

Percent of patients with
  Cancer                     38.1                15.0
  Depression                  5.8                15.3 (b)
  Surgery                    39.6                44.2

Morphine equivalents
  Index (c)                  69.1 (65.8) (b)    208.8 (331.2)
  Supplemental (d)           47.2 (249.8)        23.1 (72.8) (b)
  Combined (e)              116.3 (271.4) (b)   231.9 (361.1)

Days exposure               276.6 (284.7) (b)   326.5 (294.7)

                          Treatment group

Characteristic (mean      Morphine sulfate
[SD] or %)                CR (n = 514)

Age, yr                      56.4 (15.1)

Gender, % male               43.6

Race, %
  White                      59.9
  Black                      12.1
  Hispanic                    4.1
  Missing/Other data         23.9

Percent of patients with
  Cancer                     37.9
  Depression                 11.7
  Surgery                    42.6

Morphine equivalents
  Index (c)                 158.8 (233.5)
  Supplemental (d)           49.0 (96.7)
  Combined (e)              207.8 (276.1)

Days exposure               337.9 (353.1)

(a) CR, controlled release.
(b) Indicates significant difference (P < 0.001) from the other two
treatment groups on a given measure.
(c) Index indicates transdermal fentanyl, oxycodone CR, or morphine CR.
(d) Supplemental indicates any opioid analgesic other than one of the
three index opioids.
(e) Combined indicates the sum of index opioid dose and supplemental
opioid dose.

Fig. 2 Crude rates of constipation by treatment group. The number and
percent of constipation cases were calculated for each treatment group
in patients who received long-acting opioids for at least 3
months. Crude rates of constipation were not significantly
different (P = 0.1281, [chi square]).

Treatment Group         Percent of Patients

Transdermal fentanyl      3.7      n = 22
Oxycodone CR              6.1      n = 44
Morphine CR               5.1      n = 26

Note: Table made from bar graph.

Fig. 3 Adjusted OR for developing constipation during treatment with
long-acting opioids. OR for oxycodone CR and morphine CR was adjusted
for rescue morphine equivalents, race, and number of days of
opioid exposure, using transdermal fentanyl as the
reference (index) group.

Treatment Group         Adjusted Odds Ratio

Transdermal fentanyl    1.0     -
Oxycodone CR            1.78    (P=.0337)
Morphine CR             1.44    (P=.2242)

Note: Table made from bar graph.

Table 2. Constipation incidence density and risk ratios: summary
statistics (a)

                                           Treatment group

                                             Transdermal
                                          fentanyl (n = 601)

Number of new constipation cases                 22
Total number of exposure days (b)              166,242
Mean number of exposure days (SD)           276.6 (284.7)
Incidence density (annual rate), % (c)          4.83
Risk ratio (95% CI)                              1 (d)

                                          Treatment group

                                          Oxycodone HCI CR
                                             (n = 721)

Number of new constipation cases                 44
Total number of exposure days (b)             235,437
Mean number of exposure days (SD)           326.7 (294.7)
Incidence density (annual rate), % (c)          6.83
Risk ratio (95% CI)                       1.41 (0.85-2.36)

                                          Treatment group

                                          Morphine sulfate
                                            CR (n = 514)

Number of new constipation cases                 26
Total number of exposure days (b)              173,669
Mean number of exposure days (SD)           337.9 (353.1)
Incidence density (annual rate), % (c)          5.47
Risk ratio (95% CI)                       1.13 (0.64-2.00)

(a) CR, controlled release; CI, confidence interval.
(b) Total number of exposure days = total days' supply of transdermal
fentanyl, oxycodone HCI CR, or morphine sulfate CR during the study
period.
(c) Incidence density was calculated by dividing the number of new
constipation cases by the mean number of exposure days and multiplying
by 365.
(d) Index.

Accepted October 23, 2003.

Copyright [c] 2004 by The Southern Medical Association

0038-4348/04/9702-0129

References

(1.) Gutstein HB, Akil H. Opioid analgesics, in Hardman JG, Limbird LE, Gilman AG (eds): Goodman & Gilman's The Pharmacological Basis of Therapeutics. New York, McGraw-Hill Professional, 2001, ed 10, pp 569-619.

(2.) Allan L, Hays H, Jensen NH, et al. Randomised crossover trial of transdermal fentanyl and sustained release oral morphine for treating chronic non-cancer pain. BMJ 2001;322:1154-1158.

(3.) Donner B, Zenz M, Tryba M, et al. Direct conversion from oral morphine to transdermal fentanyl: A multicenter study in patients with cancer pain. Pain 1996;64:527-534.

(4.) Donner B, Zenz M. Transdermal fentanyl: A new step on the therapeutic ladder. Anticancer Drugs 1995;6(Suppl 3):39-43.

(5.) Radbruch L, Sabatowski R, Loick G, et al. Constipation and the use of laxatives: A comparison between transdermal fentanyl and oral morphine. Palliat Med 2000;14:111-119.

(6.) Donner B, Zenz M, Strumpf M, et al. Long-term treatment of cancer pain with transdermal fentanyl. J Pain Symptom Manage 1998;15:168-175.

(7.) Ahmedzai S, Brooks D; The TTS-Fentanyl Comparative Trial Group. Transdermal fentanyl versus sustained-release oral morphine in cancer pain: Preference, efficacy, and quality of life. J Pain Symptom Manage 1997;13:254-261.

(8.) De Luca A, Coupar IM. Insights into opioid action in the intestinal tract. Pharmacol Ther 1996;69:103-115.

(9.) Megens AA, Artois K, Vermeire J, et al. Comparison of the analgesic and intestinal effects of fentanyl and morphine in rats. J Pain Symptom Manage 1998;15:253-257.

(10.) Pasternak GW. Insights into [mu] opioid pharmacology: The role of [mu] opioid receptor subtypes. Life Sci 2001;68:2213-2219.

(11.) Caldwell JR, Hale ME, Boyd RE, et al. Treatment of osteoarthritis pain with controlled release oxycodone or fixed combination oxycodone plus acetaminophen added to nonsteroidal antiinflammatory drugs: A double blind, randomized, multicenter, placebo controlled trial. J Rheumatol 1999;26:862-869.

(12.) Maier C, Hildebrandt J, Klinger R, et al. Morphine responsiveness, efficacy and tolerability in patients with chronic non-tumor associated pain: Results of a double-blind placebo-controlled trial (MONTAS). Pain 2002;97:223-233.

(13.) Cameron JC. Constipation related to narcotic therapy: A protocol for nurses and patients. Cancer Nurs 1992;15:372-377.

(14.) Robinson CB, Fritch M, Hullett L, et al. Development of a protocol to prevent opioid-induced constipation in patients with cancer: A research utilization project. Clin J Oncol Nurs 2000;4:79-84.

Peter S. Staats, MD, Jeffrey Markowitz, DRPH, and Jeffrey Schein, DRPH, MPH

From the Departments of Anesthesiology and Critical Care Medicine and of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD; Health Data Analytics, Princeton Junction, NJ; and Janssen Pharmaceutica Products, LP, Titusville, NJ.

Reprint requests to Peter S. Staats, MD, Division of Pain Medicine, Department of Anesthesiology and Critical Care Medicine, 550 N. Broadway, Suite 301, Baltimore, MD 21205. Email: [e-mail]