FENTANYL was one of a series of opioids synthesized by Janssen Pharmaceutica in the 1950s and 1960s in an effort to produce opioid analgesics with enhanced analgesic activity and potency and fewer adverse effects compared with morphine or meperidine. [1,2]It was first used clinically as a component of neuroleptanalgesia in combination with the butyrophenone, droperidol. Between 1975 and 1981, fentanyl was adopted widely as a potent intraoperative analgesic agent with relatively few adverse effects. In small-to-moderate bolus doses (3 to 5 [micro sign]g/kg), it combined with different intravenous supplements to produce “balanced” anesthesia, whereas large doses (as much as 100 [micro sign]g/kg) were used to induce and to maintain anesthesia in critically ill patients and those undergoing cardiopulmonary bypass procedures. 
Fentanyl's popularity as an intraoperative agent relates directly to the cardiovascular stability it provides, even in critically ill patients. [6,7]But its analgesic efficacy relative to the intensity of side effects prompted much interest in its use as an analgesic agent after operation or in the intensive care unit. Investigators began by exploring alternatives to the traditional intramuscular or intravenous routes for postoperative administration to optimize the potential clinical benefits of fentanyl's physiochemical properties. This article reviews the literature related to the use of fentanyl as an analgesic in the postoperative period and in patients in the intensive care unit, and it evaluates the pharmacokinetics, pharmacodynamics, efficacy, and limitations of existing and experimental routes of administration.
Physical and Chemical Properties of Fentanyl
Fentanyl, N-(1-phenethyl-4-piperidyl) propionanilide, is structurally related to meperidine. Commercially, fentanyl is formulated as a citrate, available in a water-soluble, white crystalline powder that requires no preservatives. It has a molecular weight of 528.29 and a melting point of 148.5 to 150[degree sign]C. Each milliliter of aqueous solution contains a base of 0.05 mg fentanyl (0.0785 mg of the citrate).
The negative logarithm of the acid ionization constant of fentanyl (pKa) is 8.43. At physiologic pH, 8.5% of the compound is un-ionized in plasma and 84% is bound to erythrocytes, [small alpha, Greek]1-acidglycoprotein, and plasma albumin. The octanol-water partition coefficient at physiologic pH is 816 for fentanyl compared with 1.4 for morphine. Therefore, fentanyl is highly lipophilic, whereas morphine is hydrophilic. Multiplying this partition coefficient by the plasma-free fraction (Table 1) yields a relative potential to enter the central nervous system that is approximately 133 times larger than that of morphine. 
Optimization of the molecular configuration of fentanyl increased its potency. Fentanyl is 100 to 300 times more potent than morphine per dose, depending on the animal species. [10-12]This greater dose potency permits a low therapeutic blood concentration of approximately 0.6 to 3 ng/ml for analgesia. This, in turn, necessitates a sensitive method of assay.
Radioimmunoassay and gas liquid chromatography are the two most common methods used. The current radioimmunoassay method can measure plasma fentanyl concentrations as low as 0.06 ng/ml and was first reported in 1977. The standard curves are linear for a concentration range of 0.06-20 ng/ml, and the coefficient of variation of the assay ranges from 1-12%. [14-21]However, radioimmunoassay analysis can overestimate plasma fentanyl concentrations (fentanyl CP) by as much as 29 to 100%, limiting reliability, and thereby contributing to the observed differences in the pharmacokinetic data reported for fentanyl. 
Assay by gas liquid chromatography using either flame-ionization, nitrogen phosphorus, or mass spectrometric detection is sensitive and reproducible. With nitrogen phosphorus, the mean coefficient of variation for concentrations ranging from 0.25-10 ng/ml is 4.65%; with mass spectrometric detection, the mean coefficient of variation is 6.9% for a range of 0.2-68 ng/ml. When compared directly with the radioimmunoassay method, the gas liquid chromatography-nitrogen phosphorus method results in comparable values in the spiked control and patient samples. At the detection limit of 0.25 ng/ml, gas liquid chromatography has a coefficient of variation of 14.7%, comparable to 14.2% for radioimmunoassay. At higher concentrations, the coefficient of variation decreases to approximately 5%. This increased variability at the detection limit significantly affects pharmacokinetic analysis, because the terminal half-life for low-to-moderate doses of fentanyl (5-15 [micro sign]g/kg) is estimated using serum levels in the region of this limit. Accordingly, limitations of the assay, whether radioimmunoassay or gas liquid chromatography, must be considered in interpreting studies that profile fentanyl pharmacokinetics.
Fentanyl has both high lipid solubility and a pattern of rapid and extensive redistribution, making it an ideal agent to evaluate drug delivery systems and routes of administration other than the traditional parenteral routes. Consequently, it has been administered via intramuscular, intravenous (bolus injection, infusion, patient-controlled analgesia [PCA]), neuraxial (epidural, intrathecal), transdermal, transmucosal (oral or intranasal), and inhalational routes.
Systemic Administration: Intravenous
After an intravenous bolus, fentanyl distributes rapidly from plasma to highly vascular tissues (heart, lung, and brain). More than 80% of the injected dose leaves plasma in less than 5 min, and 98.6% leaves by 1 h. Elimination from the vascular tissue also is rapid as fentanyl redistributes to other sites, such as muscle and fat. In rats, fentanyl CPpeaks in muscle 5 min after a bolus dose, and in fat at approximately 30 min (Figure 1). Removal from muscle and fat is slower than uptake, because both tissues act as storage sites; in muscle this is because of its mass, and in fat because of the high lipid solubility of fentanyl. After initial equilibration with adipose tissue, fentanyl C (P) decreases, and then fat slowly releases the fentanyl back into the plasma. This slow release results in a lengthy elimination half-time of 3.1 to 7.9 h (Table 1). Thus, fentanyl's short duration of action after a single dose results from redistribution rather than elimination. After large or multiple smaller doses, fentanyl accumulates as a result of its long half-time, and redistribution is less effective in removing fentanyl from its site of action in the brain. [9,30]
Fentanyl is metabolized almost exclusively in the liver to norfentanyl, hydroxy-proprionyl-fentanyl, and hydroxyproprionyl-norfentanyl. The pharmacologic activity of fentanyl metabolites is unknown but is believed to be minimal. Less than 10% of fentanyl is excreted unchanged by the kidney. The total body clearance of fentanyl is high, between 8 and 21 ml [middle dot] kg-1[middle dot] min-1, and approaches that of liver blood flow, reflecting the high hepatic extraction ratio. The high lipid solubility of fentanyl contributes to a large volume of distribution (3.2-5.6 l/kg).
Several studies correlate fentanyl CPwith analgesia (the desired effect) and respiratory depression (the most dangerous side effect). However, the intensity of fentanyl's effect correlates with the drug concentration at the site of action (effect site) and not necessarily the plasma concentration. For opioids, the effect site or biophase is the opioid receptor in the brain and spinal cord. Additional time is needed for fentanyl to cross the blood-brain barrier to reach the effect site. The temporal lag between plasma concentration and the effect on the biophase is called hysteresis. A first-order rate constant (keo) characterizes the temporal aspects of equilibration between the effect-compartment concentration and the serum concentration. Thus, the half-time for equilibration t1/2keo(0.693/keo) quantifies the magnitude of the hysteresis (Table 1). Using electroencephalography to measure opioid effect, one group of investigators found a 3- to 5-min lag between increasing fentanyl CPand electroencephalography slowing during a 5-min fentanyl infusion. After the infusion was discontinued, resolution of electroencephalography changes lagged behind decreasing fentanyl CPby 10 to 20 min (Figure 2).
Different modes of administration have different degrees of hysteresis. With a rapid change in plasma concentration (e.g., after an intravenous bolus), the temporal lag will be greatest; with a slow change in concentration (e.g., with a steady continuous infusion), the lag will be smallest. Consequently, pharmacodynamic data obtained via different modes must be compared with caution.
Plasma Fentanyl Concentration and Analgesia. Most studies correlating fentanyl CPwith its analgesic and side effects have estimated plasma fentanyl from gradually changing concentrations in selected groups of patients. Data from studies limited primarily to patients receiving intravenous fentanyl for postoperative analgesia indicate a mean analgesic C (P) ranging from 0.6-3 ng/ml. [14,15,17,18,33-38]Infusion of fentanyl to achieve a steady state CPis reported in one study to produce “slight but significant analgesia” at a mean concentration of 0.6 ng/ml and “significantly greater analgesia” at a CPof 1.7 ng/ml, and, in another, an analgesic range of 1-3 ng/ml. With PCA, the mean minimum effective analgesic concentration (MEC-fentanyl CPimmediately before the patient administers the next bolus dose ) has been reported as 1.35 ng/ml, 1.54 ng/ml, and 0.63 ng/ml. Thus, mean MEC values range from 0.6-1.54 ng/ml, whereas values for individual patients range from 0.2-8.0 ng/ml with a log-normal distribution. 
Studies correlating fentanyl CPwith analgesic effect via visual analog pain scale ([VAS] 0 = no pain, 10 = maximum pain) scores report that the mean fentanyl CPof 0.3 to 0.7 ng/ml and 0.5 to 1.2 ng/ml during PCA correlate with VAS scores at rest of 3 or 4 and 2 to 4, respectively. Scores of 1 to 3 are associated with a CPof 1 or 2 ng/ml in different postoperative patient populations (thoracotomy, [17,18]cesarean section, knee surgery ) treated by a bolus dose plus infusion of fentanyl. However, measurements of the CPassociated with effective analgesia often are obtained while patients are at rest; at a similar CP, VAS scores markedly increase with movement or coughing. 
The observed variability in the analgesic CPreported for fentanyl in large part is caused by differences in study design and in individual pharmacodynamic responses. Analgesic requirements of individual patients and different surgical populations vary over a sixfold range for fentanyl and other opioids. [15,37,40]With respect to study design, the residual presence of anesthetic drugs and possible coadministration of central nervous system depressants affect the relation between doses or the C (P) of fentanyl and the intensity of analgesia and side effects. The degree of drug interaction also varies by study design. The types of surgical procedure also alter the degree of postoperative pain, and thereby the analgesic requirement: With a similar study design, we would expect a higher analgesic requirement in patients undergoing thoracotomy than hysterectomy. Different measurements of analgesic effect are used, including descriptive terms, MEC, CPassociated with 50% reduction in pain intensity, and VAS score. [17-19,38]The timing of blood sampling also differs: Some investigators sample at predetermined intervals, correlating these results with analgesic effect, whereas others measuring MEC sample just before the patient administers the next bolus. [15,36,37]
Intravenously administered fentanyl produces effective analgesia in patients after operation at CPvalues ranging from 0.6-3.0 ng/ml. Pain control at rest is satisfactory within this range, but the analgesic effect diminishes with movement or coughing, suggesting that a higher CPmay be required if analgesia is intended to promote either of these responses.
Plasma Fentanyl Concentration and Respiratory Depression. Studies investigating the connection between fentanyl CPand ventilatory effect show a concentration-effect relation. [33,41], [double dagger] With intravenous bolus elimination, a CPof 3 or 4 ng/ml produces a 50% decrease in the slope of minute ventilation (VE) versus end-tidal carbon dioxide concentration. [20,21,42]However, these measured plasma concentrations may not reflect effect-site concentrations of fentanyl, as a result of the hysteresis between these values with bolus administration. Other studies using prolonged infusions or loading dose/maintenance infusions thus provide a better reflection of the steady state concentration-effect relations for fentanyl, and these report clinically significant respiratory depression in both patients undergoing surgery [1,34,43]and volunteers at a steady state CPof 2 ng/ml or more. For this review, we define clinically significant respiratory depression as a requirement for intervention with naloxone administration, resuscitation, or cessation of fentanyl treatment.
Fentanyl-induced respiratory depression has been measured primarily by assessing the ventilatory response to carbon dioxide using the carbon dioxide rebreathing technique. Although an altered carbon dioxide response may indicate depression of central respiratory control, this approach may be impractical in patients after operation because it relies on patient cooperation and is significantly affected by alertness and arousal, conditions that are likely to vary among patients. Continuous measurement of ventilation therefore is preferable, particularly to detect the occurrence of apnea, hypopnea, slow respiratory rate, or hypoxemia, the development of which will not be detected by discrete measurement. 
Respiratory inductive plethysmography and pulse oximetry have been used to monitor the occurrence of episodes of apnea (tidal volume < 100 ml for > 15 s in adults) and slow respiratory rate (< 8 breaths/min for > 5 min). [45-47]Using these methods, Sandler et al. reported a baseline occurrence of three apnea episodes per hour before operation in patients undergoing thoracotomy. In the postoperative period, a steady state CPof 1 or 2 ng/ml was associated with VAS pain scores of 2 at rest, a moderate increase in apneic episodes and slow respiratory rates, and the partial pressure of carbon dioxide arterial blood levels of 47-49 mmHg. This degree of respiratory depression did not require intervention.
There is a direct concentration-effect relation between the CPfentanyl and respiratory depression. Plasma concentrations greater than 2 ng/ml are associated with clinically significant respiratory depression. However, the degree of respiratory depression is affected by various factors, including the types of surgical population, level of noxious stimulation, age, and individual pharmacodynamic responses. Thus, a threshold greater than 2 ng/ml should serve primarily as a guideline for clinicians.
Therapeutic Window. The therapeutic window for fentanyl analgesia is the range between the minimally effective analgesic concentration and that associated with respiratory depression. Studies in volunteers allow the complex relations among analgesic effect, side effects, and fentanyl C (P) to be evaluated under highly controlled conditions. In volunteers, as in patients, the magnitude of analgesia and respiratory depression has a direct relation to the fentanyl CP. The lowest concentration producing slight, but measurable analgesia, without having significant ventilatory effect in volunteers and patients is 0.6 ng/ml. At plasma concentrations at which pain intensity decreases by 50%(CPassociated with 50% reduction in pain intensity = 1.4 ng/ml), fentanyl decreases the slope of the VEversus end-tidal carbon dioxide concentration curve in volunteers by 33% and VEby 12%. At a CPof 3 ng/ml, fentanyl produces profound analgesia and decreases this slope by 54% and VEby 23%. Thus, the therapeutic margin in volunteers correlates reasonably well with that just described for patients after operation.
There is a direct concentration-effect relation between fentanyl C (P) and analgesia and respiratory depression. In volunteers and patients, the range of fentanyl CPproviding analgesia without clinically significant respiratory depression is 0.6-2 ng/ml. Factors including type of surgical procedure, surgical population (e.g., elderly patients), interaction with other central depressive drugs, and individual pharmacodynamic and pharmacokinetic differences can markedly influence this window and should be considered when the suggested therapeutic range is applied.
Modes of Administration. Fentanyl can be administered intravenously for postoperative analgesia using a loading (bolus) dose with a continuous fixed or variable infusion, a fixed background infusion with PCA, or PCA alone.
Continuous Background Intravenous Fentanyl Infusion. Continuous intravenous infusions of fentanyl have been used to provide postoperative analgesia after abdominal, [14,35,49]peripheral orthopedic, [35,39,50]and major spinal surgery ; thoracotomy [17,18,50,52]; and cesarean section delivery (Table 2).
Dose Requirements. An intravenous bolus of fentanyl (1 or 2 [micro sign]g/kg) usually is administered before the start of infusion. If variable, the infusion rate is 1 or 2 [micro sign]g [middle dot] kg-1[middle dot] h-1(Table 2) and may be adjusted upward or downward as required by fluctuations in analgesic requirements or increasing side effects. Before the infusion rate is increased, small bolus doses of fentanyl are administered to increase the CPrapidly. [17,18,38,39]If the infusion rate is fixed, a supplementary analgesic technique, either bolus or PCA doses of nonsteroidal antiinflammatory drugs, fentanyl, or morphine (Table 2), is used to meet the therapeutic demand. [35,50,51]
Analgesic Efficacy. Infusion of fentanyl, especially at rates of 1.5-2.5 [micro sign]g [middle dot] kg-1[middle dot] h-1, can provide good-to-excellent postoperative analgesia (Table 2). At rest, the quality of analgesia remains stable; with movement (ambulation, coughing), it decreases significantly, even with higher infusion rates. [39,52]The fentanyl CPrelates directly to the infusion rate, with good analgesia at rest associated with concentrations ranging from 0.5-2.3 ng/ml.
Side Effects. Nonrespiratory side effects can occur. The incidence of nausea and vomiting after fentanyl infusion varies from 20-60%. Pruritus occurs in 0-30% of patients, and urinary retention occurs in 40-45%. The latter values may over- or underestimate the actual incidence of urinary retention, because many studies use postoperative indwelling urinary catheters that preclude measurement of urinary retention.
Respiratory depression is common after fentanyl infusion, but most events are not significant. Only three studies [17,50,51]have reported clinically significant respiratory depression. Comparing these three with other studies (Table 2) reveals similar infusion doses, types of surgery, and other factors (e.g., systemic morphine supplementation), making it unclear why the incidence of respiratory depression differed significantly. The methods of detection and measurement of respiratory depression include intermittent or continuous measurement of respiratory rate, pulse oximetry, respiratory inductive plethysmography, and intermittent arterial blood sampling.
Continuous intravenous fentanyl infusion provides good-to-excellent analgesia (particularly at rest) at doses of 1 or 2 [micro sign]g [middle dot] kg-1[middle dot] h-1. Naturally occurring variations in postoperative analgesic requirements can be managed by adjusting the fentanyl infusion rate upward or downward, as needed, assuming a variable infusion technique, or by parenteral administration of bolus doses of opioid or nonsteroidal antiinflammatory drugs to supplement a fixed infusion.
Continuous Background Intravenous Infusion with Patient-Controlled Analgesia. A background low-dose intravenous infusion of fentanyl may be combined with PCA to provide satisfactory analgesia with potentially fewer adverse effects. [15,36,53-58]
Dose Requirements. Patient-controlled analgesia bolus doses typically range from 7-50 [micro sign]g. Background infusion rates may be fixed, ranging from 4-60 [micro sign]g/h, or be variable, adjusted up and down according to clinical criteria [54,55](Table 3). Generally, the larger the background infusion rate, the smaller the PCA bolus dose. Lockout intervals (minimum time period between doses) range from “on demand”(i.e., no lockout) to 15 min, the most common interval being 1-5 min (Table 3).
No study directly compares the use of intravenous fentanyl infusion with and without fentanyl PCA. However, examination of the data in Table 2and Table 3reveals a smaller dose requirement for continuous infusion plus PCA than for infusion alone, despite variability in infusion rate, PCA dose, and lockout interval. The type of surgery also influences the dose requirement; that is, thoracotomy generates higher requirements than does orthopedic or lower abdominal surgery. 
Analgesic Efficacy. The technique of background infusion plus PCA produces excellent postoperative pain relief for abdominal, [15,37,54,56,57]orthopedic and peripheral vascular surgery and for thoracotomy. [55,58]Consistent with the primary fentanyl infusion studies, pain has been measured in patients at rest [15,58]and increased markedly with movement. 
Studies of the concentration-effect relation with this technique have shown a MEC value for fentanyl for abdominal and orthopedic surgery that varies from 0.63-1.54 ng/ml. [15,36,37]
Side Effects. Nonrespiratory side effects can occur, but data regarding the incidence of adverse effects with this technique are limited. Nausea and vomiting are estimated to occur in approximately 30-40% of patients. Only two studies have investigated the incidence of pruritus, reporting a range of 7-13%, [55,58]and none have studied the occurrence of urinary retention.
Of nine studies, only five report the incidence of respiratory depression, and none show evidence of clinically significant respiratory depression requiring treatment.
Compared with continuous infusion alone, the use of a background fentanyl infusion with PCA fentanyl provides excellent postoperative analgesia, with a lower total dose consumption. The incidence of side effects with the two techniques is difficult to compare because of the limited data published for background infusion.
Patient-Controlled Analgesia. Fentanyl is rarely used alone for PCA, most likely because of the widespread belief in its brief duration of action. The opioids most commonly administered are morphine and meperidine. Consequently, only a few studies have compared the efficacy and safety of PCA fentanyl with those of other opioids, [60-62]but other investigators have used PCA fentanyl only as a control in clinical trials comparing it with other analgesic techniques [16,19,63-67](Table 4).
Dose Requirements. Consistent with other opioids used for PCA, [60,62]fentanyl dose requirements vary widely. [36,68]Bolus doses range from 20-50 [micro sign]g with lockout intervals from “on demand” to 10 min. Theoretically, the lockout interval should relate to the time from drug administered to peak effect so that patients can experience the full effect of a dose before receiving a subsequent dose. Because of fentanyl's short latency to peak effect, a lockout interval of 5 or 6 min is reasonable.
Maximum hourly PCA dosage varies from 120-600 [micro sign]g/h, and mean hourly requirements vary from 48-83 [micro sign]g/h (Table 4), less than those for the continuous infusion technique (Table 2). Despite fentanyl's brief duration of action, patients typically require only one to three doses/h and rarely administer more than two bolus doses/h. [63,69,70]At bolus doses of 20-50 [micro sign]g, fentanyl therefore might be suitable for PCA use.
Analgesic Efficacy. Good analgesia can be achieved with PCA fentanyl alone, with efficacy comparable to that of morphine and meperidine. [61,62]Only two studies, [60,67]both using small bolus doses (20-25 [micro sign]g) and a long lockout interval (10 min), report inadequate analgesia.
No study has directly compared the influence of a background infusion on the efficacy of intravenous fentanyl delivered by PCA. Comparing individual studies in Table 3and Table 4suggests that PCA fentanyl alone produces similarly effective analgesia with similar dose requirements as PCA with a background infusion. Most studies of other opioids fail to show any benefit to adding a background infusion to PCA. [71-74]Furthermore, the use of such an infusion increases opioid requirements [71-73,75]and is associated with an increased risk for severe respiratory depression. [76-78]Therefore, it may be prudent to avoid using background infusions with PCA fentanyl.
Depending on the dose and lockout interval set for the PCA device, relatively effective analgesic fentanyl CPcan be achieved and maintained with PCA fentanyl alone. In an unblinded trial of 11 patients undergoing upper abdominal surgery, a bolus dose of 20 [micro sign]g with a lockout interval of 6 min resulted in a mean fentanyl CPof 1.4 +/− 0.7 ng/ml (mean +/− SD) 12 h after operation, which decreased to 0.5 +/− 0.2 ng/ml at 48 h. Mean VAS pain scores at rest at 12 and 48 h were acceptable at 4 and 3, respectively. Comparing PCA fentanyl administered through the epidural route with PCA intravenous fentanyl in patients undergoing lower limb orthopedic or abdominal surgery, Glass et al. conducted a randomized, double-blind, crossover trial using the same 20-[micro sign]g bolus dose and 6-min lockout interval. Fentanyl CPfor the first 6 h in the intravenous PCA group ranged from 0.2-0.4 ng/ml, resulting in mean VAS scores for this period of 2 to 4 at rest (i.e., moderate-to-good analgesia).
Side Effects. Only a few studies have reported the incidence of nonrespiratory side effects with PCA fentanyl alone. Nausea and vomiting occur in 20-60% of patients and pruritus occurs in 0-40%.
There are no reports of clinically significant respiratory depression with PCA fentanyl alone. However, all these studies thus far monitored respiratory depression solely by respiratory rate, which correlates poorly with ventilatory insufficiency. 
Patient-controlled analgesia fentanyl provides analgesia comparable to that of other intravenous modes of administration. Despite fentanyl's short duration of action, most patients require only one to three bolus doses/h. The addition of a background infusion offers no benefit to the quality of analgesia and potentially increases the risk of respiratory depression. Compared with continuous infusion, average dose consumption is less with PCA alone.
Systemic Administration: Transdermal
Transdermal delivery of fentanyl has been investigated extensively. This modality is simple, noninvasive, and allows continuous release of fentanyl into the systemic circulation. The major barrier to the entry of transdermally administered drug into the systemic circulation is the stratum corneum of the epidermis. This layer of skin has a “brick-and-mortar” arrangement of kiratin-rich cells embedded in a lipid matrix arranged in broad sheets forming multiple layers. However, fentanyl's lipid-soluble properties allow it to diffuse through the stratum corneum via the intercellular lipid medium. 
Passive (Conventional) Transdermal Fentanyl Administration
Permeability of the stratum corneum may be affected by various factors, including body site, skin temperature, skin damage, ethnic group, or age. To ensure a predictable rate of drug transfer, the transdermal delivery system minimizes the influence of skin in transfer by incorporating a rate-controlling membrane more impermeable than skin.
The Therapeutic Transdermal System (TTS; ALZA Corp., Palo Alto, CA) uses the membrane permeation model. This transdermal fentanyl patch is available in four sizes and provides sustained release of fentanyl at rates of approximately 25, 50, 75, and 100 [micro sign]g/h for periods of 48-72 h. The patch is attached to the skin by a contact adhesive, adjacent to which is a microporous membrane that controls the rate at which fentanyl is transferred from the drug reservoir to the skin (Figure 3). The reservoir is a shallow compartment with a gel matrix containing as much as 10 mg fentanyl, intended to provide a sufficiently high concentration gradient for diffusion across the skin. To prevent escape of the fentanyl matrix into the environment, the reservoir has a backing.
An important feature of the TTS design is that it takes advantage of the substantial capacity of the skin layers to act as a secondary reservoir. The presence of skin depot has several implications: It dampens the fluctuations of fentanyl effect, needs to be reasonably filled before significant vascular absorption occurs, and contributes to a prolonged residual fentanyl CPafter patch removal. The amount of fentanyl remaining within the system and skin depot after removal of the patch is substantial: At the end of a 24-h period with a TTS fentanyl patch releasing drug at the rate of 100 [micro sign]g/h, 1.07 +/− 0.43 mg fentanyl (approximately 30% of the total delivered dose from the patch) remains in the skin depot. 
Pharmacokinetics. Pharmacokinetic studies have examined absorption, plateau systemic concentrations, time to peak concentration, and apparent elimination half-life after removal of the TTS fentanyl patch. [24,45,49,50,82-89]There are two distinct phases of fentanyl absorption after a 24-h application of the TTS patch: An initial phase with rapid skin absorption of the drug from the contact adhesive because of the large concentration gradient between the patch and the skin reservoir, and a plateau phase with sustained release of drug from the reservoir. Because of the presence of the skin reservoir, CPpeaks an average of 24 h (range, 14-28 h) after the patch is applied. During a steady state infusion, the plateau systemic concentration is proportional to the “infusion rate” of the TTS fentanyl patch (CP= infusion rate/clearance). The plateau systemic concentration values for the TTS patch doses of 50, 75, and 100 [micro sign]g/h are approximately 1, 1.5, and 2 ng/ml, respectively, well within the analgesic range for fentanyl. However, there is large interpatient variability in peak systemic concentration: Peak fentanyl CPwithin 24 h of application of the 75-[micro sign]g/h TTS fentanyl patch ranges from 1-5.5 ng/ml. [24,82,83,90]The actual infusion rate for the 75-[micro sign]g/h patch, estimated from the formula (initial fentanyl amount - residual fentanyl amount)/duration of system application, ranges from 60-130 [micro sign]g/h. 
When a TTS fentanyl patch is kept in situ for 72 h, fentanyl CPtends to decrease after 48 h. [45,88,89]After the patch is removed, fentanyl CPdecreases slowly because of continued absorption from the fentanyl skin depot. The apparent terminal half-time ranges from 16-25 h. Mean bioavailability is 92 +/− 33%, which is determined by calculating the total systemic absorption of fentanyl and measuring the total fentanyl administration (loss from the patch). 
Dose Requirements. Dose requirements using TTS fentanyl are difficult to match to individual patients or types of surgery, because the TTS fentanyl device is a constant-rate infusion system that requires a long time to reach a plateau CP. In general, the patch size is selected empirically to match the magnitude of postoperative pain associated with the surgical procedure. Severe postoperative pain after major abdominal surgery usually is managed with the TTS patch releasing a 100-[micro sign]g/h dose, whereas mild to moderate pain after lower abdominal surgery warrants the patch releasing the 50-[micro sign]g/h dose. Because the release of fentanyl from the TTS patch is similar to that from a fixed intravenous infusion, systemic opioids (morphine, fentanyl, piritramide) are used to supplement individual analgesic requirements as needed. Because of the slow attainment of an analgesic CP, patches usually are applied before operation at varying time intervals, including immediately before surgery and at 0.5, 1, , 2, [45,49,50,79,83,86,92]and 8 h after operation. The effect of a single patch application lasts 24-72 h. [45,87,90-92]
Analgesic Efficacy. The TTS fentanyl system provides a steady release of fentanyl to the systemic circulation without the flexibility of dose adjustment. This may result in poor matching to the rapidly changing intensity of postoperative pain. Thus, parenteral opioids are necessary to supplement analgesia and have been administered in all studies evaluating the transdermal patch to treat acute postoperative pain. With supplementation, TTS fentanyl produces significantly better postoperative analgesia than supplemental opioid alone, independent of the route of administration, [45,79,84-86,93]and TTS fentanyl significantly reduces the supplemental dose requirement. [45,58,79,84-86,91]
Side Effects. The incidence of nausea and vomiting with TTS ranges from 10-90%. In most studies, it is 50-60%. These values are comparable to those associated with intravenous modalities and the control groups with “rescue” opioid (morphine and meperidine) bolus administration. [45,91,93]Pruritus occurs in 4-39% of patients but typically does not require treatment. Urinary retention has been measured in only a few studies at a rate of 3-27%. Skin reaction (erythema) occurs in 20-60% of patients, but it is rarely severe enough to warrant early removal of the TTS fentanyl patch and seldom persists longer than 24 h after the patch is removed.
(Table 5) shows the incidence, characteristics, and severity of respiratory depression reported by TTS fentanyl studies.
Premarketing evaluation of the safety and effectiveness of the TTS fentanyl system (Duragesic; Janssen Pharmaceutica, NJ) to treat postoperative pain found a high incidence of hypoventilation, resulting, in some cases, in death. A review of the literature on the use of TTS fentanyl patches reveals a high incidence of clinically significant respiratory depression, confirming risk (Table 5). Without further improvement in the mode of delivery or restriction of its use in closely monitored settings, the transdermal fentanyl patch delivery system cannot be recommended to treat acute pain of any origin. The Food and Drug Administration has made specific recommendations that the TTS fentanyl system should not be used to treat acute pain.
Slow onset time, inability to adjust dose during the period of application, persistent CP, and a high incidence of respiratory depression make the transdermal fentanyl patch delivery system undesirable to treat acute pain of any origin.
Active (Iontophoretic) Transdermal Fentanyl System
To overcome the resistance to drug absorption of the stratum corneum layer of the epidermis, new methods of enhancing transdermal drug penetration and absorption are being investigated.
Iontophoresis is one method to enhance transdermal drug delivery. The system consists of a skin delivery electrode, a skin current returning electrode, and an electric power source. By applying an external electric field, electrically charged components of drug are propelled through the skin. This approach for noninvasive administration has been used to deliver corticosteroids to treat joint pain, [96,97]local anesthetics (lidocaine) for analgesia for minor surgical procedures, and clinically significant doses of morphine. 
Iontophoretic administration of fentanyl has been studied in volunteers. Fentanyl (3 mg/ml) was applied for 2 h at 1 and 2 mA of current, respectively, on two occasions 2 weeks apart. Mean times to initial detection of fentanyl in the systemic circulation for the 1-mA and 2-mA applications were 33 min (range, 10-50 min) and 19 min (range, 15-20 min), respectively. Mean times to 0.5 [micro sign]g/ml CPwere 92 and 36 min, and the mean times to maximum concentrations were 122 and 119 min. Maximum C (P) was 0.76 +/− 0.23 ng/ml (mean +/− SD) with the 1-mA application and was approximately double that with the 2-mA application (Figure 4). The mean terminal half-life was similar for both the 1-mA and the 2-mA application, 354 +/− 100 and 413 +/− 106 min, respectively. The results indicate a significant relation between charge and the administered fentanyl dose. Fentanyl CPincreased throughout the 2-h delivery time.
Adverse events, including pruritus, transient hemoglobin desaturation, and hypoxemia occurred in several volunteers. Erythema was observed at the site of the dispersive pad but resolved without treatment within 24 h. No studies have described iontophoretic fentanyl use for postoperative analgesia.
Iontophoresis can be used to deliver clinically significant doses of fentanyl. A charge-dose relation has been documented for fentanyl administration by this method and could permit future development of noninvasive PCA using fentanyl. The terminal half-life estimated from the iontophoresis transdermal method is less than that from passive transdermal delivery of fentanyl, suggesting that there may be minimal skin depot effect using iontophoresis. Further research into the pharmacokinetics and analgesic efficacy of this experimental technique is required before its introduction into clinical use.
Systemic Administration: Transmucosal
The application of opioids to mucosal surfaces to achieve an effect is not a new concept. For centuries, this route has been used to self-administer opium. In current clinical settings, fentanyl had been delivered across oral and nasal mucosal membranes to achieve effective analgesia. [101,102]
Oral Transmucosal Administration
Oral transmucosal fentanyl citrate (OTFC) incorporates fentanyl citrate in a candy mixture shaped into a lozenge on a stick. The citrate salt of fentanyl is soluble in both water and several candy matrices, and it is resistant to heat, which allows it to be incorporated into a buccal lozenge system. Oral transmucosal fentanyl citrate has been assessed thoroughly as a preoperative medication in children, [103-107]and it has produced dose-dependent increases in sedation and analgesia in adult volunteers. 
Pharmacokinetics. With oral transmucosal administration, fentanyl can be absorbed directly into systemic circulation through the oral mucosa or swallowed in saliva and absorbed through the gastrointestinal tract. Fentanyl absorbed through the latter route undergoes moderate first-pass extraction in the liver. Thus, the amount of saliva swallowed before adequate exposure of fentanyl to mucosal surfaces is critical in overall absorption and probably accounts for much of the interpatient variability associated with OTFC delivery. 
The pharmacokinetics of OTFC have been determined in volunteers. Within doses ranging from 200-1,600 [micro sign]g, OTFC exhibits dose-proportional pharmacokinetics. [Section] In a study comparing oral administration of fentanyl with OTFC, volunteers were given the same fentanyl dose (15 [micro sign]g/kg) orally or transmucosally. The OTFC was placed in the buccal pouch and consumed in 15 min. The peak fentanyl CPwas two times greater with OTFC than with oral administration (3 +/− 1 vs. 1.6 +/− 0.6 ng/ml) and was achieved earlier (23 +/− 3.4 vs. 101.3 +/− 48.8 min). Bioavailability with OTFC was 46-52%, , [double vertical bar] compared with 32% with oral administration. Fentanyl CPdecreased to less than 1 ng/ml within 75-135 min of initial mucosal exposure to OTFC. The speed of this decrease in concentration and the comparability of terminal half-life values after intravenous and OTFC administration suggest that a fentanyl depot does not develop in the oral mucosa. 
Dose Requirements. Only a few studies have investigated the use of OTFC in postoperative settings. [101,110]In a double-blind, randomized controlled trial, OTFC (7-10 [micro sign]g/kg) was administered three times at 4-h intervals on the first postoperative day in adult patients undergoing hip or knee arthroplasty. Patient-controlled analgesia morphine supplementation was available to patients in both OTFC and control groups, as needed. The OTFC dose per treatment averaged 9.7 [micro sign]g/kg. Similar VAS pain scores (VAS 2 or 3) were achieved with the two modalities, at approximately half the supplemental PCA dose in the OTFC group. Another double-blind, randomized trial conducted in patients undergoing lower abdominal surgery showed that a single dose of 800 [micro sign]g OTFC provided better and more sustained analgesia than a dose of 200 [micro sign]g, with an onset and duration of effect similar to those achieved with a single intravenous bolus of 10 mg morphine.
Analgesic Efficacy. The median time to onset of analgesia with OTFC is approximately 4 min. The duration of effect varies by patient, and doses used, ranging from 159 +/− 91 min with a dose of 200 [micro sign]g to 220 +/− 112 min with a dose of 800 [micro sign]g. The quality of analgesia is good, as indicated by reported VAS pain scores of 2 or 3. A single 800-[micro sign]g dose of OTFC results in an onset, duration, and quality of analgesia comparable to those of a single, intravenous 10-mg bolus of morphine in patients experiencing mild to moderate postoperative pain.
Side Effects. The incidence of adverse effects with OTFC supplemented by PCA morphine appears to be comparable to that with PCA morphine alone. Pruritus, nausea, and vomiting have been reported in 20%, 40%, and 13% of patients, respectively, using OTFC plus PCA morphine and 8%, 46%, and 15% with PCA morphine alone. Similarly, episodes of hemoglobin oxygen desaturation to less then 90% with OTFC and PCA morphine alone are 13% and 31%, respectively (P value not significant).
Oral transmucosal fentanyl citrate appears to provide analgesia of rapid onset and medium duration, comparable to that achieved with an intravenous bolus of morphine. Despite a reduction in supplemental morphine consumption, the use of OTFC did not decrease the incidence of adverse events. With limited data on the use of OTFC in the postoperative period, its role as a useful postoperative analgesic technique is not well defined.
The surface area of the nasal cavity in a normal adult is approximately 180 cm2, and the entire cavity is highly vascularized, with blood flow of 40 ml [middle dot] min-1[middle dot] 100 g-1of tissue. Although the pharmacokinetics of fentanyl via the intranasal route have not been evaluated, agents that are known to be lipophilic and have a low molecular weight (e.g., propranolol) produce serum concentrations similar to those achieved with intravenous administration. Thus, intranasal fentanyl might exhibit similar pharmacokinetic profiles.
Dose Requirements. The only two studies to date to establish a dose requirement for intranasal fentanyl compared the effects of intranasal and intravenous fentanyl in the 60 min immediately after surgery. [102,113]Fentanyl was administered with a metered device, with each spray delivering 4.5 [micro sign]g fentanyl. A dose of six sprays (27 [micro sign]g fentanyl) was delivered after various procedures and repeated every 5 min until patients were free of pain or refused any further analgesic. The same dose regimen was administered intravenously to the control group. An average of 3.9 nasal doses (range, 1-9 doses) of 27 [micro sign]g each resulted in excellent VAS pain scores at rest in the immediate postoperative period. The mean dose requirement of 106 +/− 60 [micro sign]g did not differ from that in the intravenous group (99 +/− 60 [micro sign]g).
Analgesic Efficacy. Data from the same studies indicate mean times to onset and peak analgesic effect with intranasal fentanyl of 16.0 +/− 12.6 min and 26.3 +/− 15 min (mean +/− SD), respectively, which are both slower than the times achieved with intravenous administration (10.8 +/− 9 min and 20.2 +/− 12 min, respectively). In part, the slow onset time with intranasal administration may have resulted from the study design with small, incremental doses. The maximum pain-relieving effect was the same with both techniques.
Side Effects. Intranasal fentanyl did not irritate the nasal mucosa and produced only minimal respiratory effects (mild hemoglobin desaturation in 1 of 53 patients) and an incidence of nausea, vomiting, and euphoria less than 1%.
The value of intranasal administration of fentanyl for postoperative analgesia needs to be further defined. Based on only two studies, it appears that intranasal delivery of fentanyl can produce analgesia similar to that achieved by intravenous administration. However, use of a low dose slows onset time.
Systemic Administration: Transpulmonary
Transpulmonary (inhalational) administration of medication produces rapid, effective drug delivery as a result of the thin alveolar-blood barrier, high tissue perfusion, and enormous surface area of the lungs. Delivery of morphine through the pulmonary system has proved effective. [114,115]Fentanyl also can produce postoperative analgesia if administered as a nebulized aerosol. [116,117]To overcome fentanyl's short duration of action, a liposome-encapsulated drug carrier system has been developed. Liposomes are microscopic vesicles composed of an aqueous compartment surrounded by a phospholipid bilayer that acts as a permeable barrier to entrap molecules. Incorporation of a drug within a liposome provides a controlled, sustained release system.
Data on the pharmacokinetics of transpulmonary fentanyl are limited. Inhalation of 300 [micro sign]g fentanyl from the nebulizer produces a peak CPof 0.4 ng/ml at 2 min and a plateau CPof 0.1 ng/ml at 15 min. With inhalation of 100 [micro sign]g, fentanyl CPremains stable at a concentration close to the detection limit of 0.02 ng/ml. 
In a study comparing nebulizer and intravenous administration of fentanyl, delivery of 2,000 [micro sign]g of a nebulized mixture of free (50%) and liposomal-encapsulated (50%) fentanyl (FLEF) to volunteers resulted in a peak CPof 1.15 ng/ml at 22 min. Although venous blood was sampled in this study, this should be of little consequence because, at 22 min, arterial and venous concentrations differ minimally. One important feature of the FLEF was that the CPdecreased slowly after the single 2,000-[micro sign]g dose: At 8 and 24 h after inhalation, fentanyl CPvalues were 0.25 +/− 0.14 ng/ml and 0.12 +/− 0.16 ng/ml, respectively.
In a subsequent study, five doses of 4,000 [micro sign]g FLEF were administered at 12-h intervals. The time to reach the peak concentration after each administration ranged from 12.5-19.2 min, and the fentanyl CPwas maintained within the analgesic therapeutic concentration; that is, 0.6-3 ng/ml (Table 6). The bioavailability of inhaled FLEF is 12-20%, [119,120]which is consistent with the bioavailability of most drugs administered via the pulmonary system (10-20%). 
Dose Requirements. A dose of 300 [micro sign]g fentanyl administered via an oxygen-driven nebulizer significantly decrease pain after various surgical procedures. It also significantly increases the median time to the first supplemental dose of parenteral morphine relative to control (190 vs. 9 min, respectively). However, in another study in which a single nebulized dose of 3 ml fentanyl citrate solution of different concentrations (100, 250, and 500 [micro sign]g/ml) administered progressively in 9 min, only the highest dose (500 [micro sign]g/ml) resulted in a moderate analgesic response within 5 min of inhalation. The differences in these results probably can be accounted for by differences in study design, nebulizer administration technique, and surgical population. No studies have reported the dose requirements for inhaled liposomal-encapsulated fentanyl in patients after operation.
Analgesic Efficacy. After inhalation of nebulized fentanyl, moderate analgesia is achieved in 5 min, the time when CPpeaks. However, this analgesic effect only 2 h. The FLEF mixture has the potential to prolong the analgesia, as indicated by the presence of a therapeutic concentration 12 h after inhalation of a single dose of 4,000 [micro sign]g. 
Side Effects. No clinically significant respiratory depression or evidence of respiratory tract irritation has been reported in the few patients studied thus far, nor is there any significant difference in nausea and drowsiness relative to controls.
Inhalation of fentanyl offers an easy, noninvasive route of administration. Onset of effect is rapid after nebulizer administration of fentanyl at a high dose (1,500 [micro sign]g). However, the duration of action with this technique is too brief for routine clinical use. The liposome-encapsulated method significantly prolongs the effect of fentanyl, but it slows the onset of analgesia. Additional study is required to determine the safety and efficacy of transpulmonary fentanyl administration for postoperative analgesia.
Epidural and intrathecal administration of fentanyl are long-established routes for intraoperative anesthesia and postoperative analgesia. The pharmacokinetics of epidural delivery have been well-studied, but relatively little is known about the systemic kinetics of intrathecal fentanyl.
The main routes of distribution after administration of fentanyl into the epidural space include (1) movement across the meninges into the cerebrospinal fluid (CSF);(2) movement from the CSF into the opioid receptor or other nonspecific binding site in the spinal cord;(3) rostral migration via the CSF to supraspinal sites;(4) vascular absorption in the epidural or spinal vascular system; and (5) uptake into epidural fat.
Factors that affect dural penetration include lipid solubility, molecular weight, molecular shape, and the degree of molecule ionization. [122-124]Lipid solubility, as assessed by the octanol-buffer partition coefficient, correlates with the permeability coefficient in a non-linear manner. The optimal octanol-buffer distribution coefficient that results in maximal meningeal permeability lies between 129 (alfentanil) and 560 (bupivacaine). This relation between lipid solubility and the meningeal permeability coefficient of a drug can be explained by the dual nature of the arachnoid mater, which is the principal meningeal permeability barrier. To cross the arachnoid mater, a drug molecule must diffuse through the hydrophilic domain (e.g., extracellular and intracellular fluid) and hydrophobic domain (e.g., cell membrane lipids). Lipophilic drugs readily dissolve in the lipophilic component of the arachnoid mater and thus cross the region easily. The hydrophilic zone is more difficult for these drugs to penetrate, creating the rate-limiting factor for diffusion via the arachnoid membrane. As a result, membrane permeability is highest in the opioids having intermediate lipid solubility (e.g., fentanyl).
Because of its high octanol-buffer partition coefficient, fentanyl also has high vascular permeability and moves as easily into the intravascular compartment as into the subarachnoid compartment. The extent of vascular absorption is influenced by various factors, including dose administered, the mode of administration (bolus vs. infusion), and, possibly, concurrent use of epinephrine.
After epidural bolus administration, systemic absorption of fentanyl increases as the bolus dose increases. For example, administration of a bolus of 30-70 [micro sign]g ([tilde operator] 0.5-1 [micro sign]g/kg) results in a peak fentanyl CPless than 0.2 ng/ml in the first 30 min, [125,126]which is much less than the range for MEC. However, within 10-30 min of an epidural bolus of 100-200 [micro sign]g ([tilde operator] 1.5-3 g/kg), the plateau systemic concentration is 0.3-0.86 ng/ml, [41,126-128]; that is, it is within the range of the MEC.
With continuous epidural infusion of fentanyl, clearance from the blood determines the blood concentration at steady state. Continuous infusion of doses resulting in good analgesia produces significant fentanyl CP(1 or 2 ng/ml) after several hours of administration. [18,19,38,39,129,130]Studies comparing therapeutic doses of epidural and intravenous fentanyl for postoperative pain relief show similar fentanyl CPvalues at equal and equianalgesic doses (Table 7A). [18,19,38,39]With thoracic epidural administration (i.e., thoracotomy), dose requirements [17,131]and CPare lower compared with lumbar epidural administration, but systemic absorption remains significant. There are no published data on the pharmacokinetics of fentanyl in CSF with continuous epidural infusion or repeated bolus administration (patient-controlled epidural analgesia [PCEA]).
Adding 1:300,000 epinephrine to a continuous thoracic epidural infusion of fentanyl significantly reduces the fentanyl CPand, relative to intravenous delivery, the equianalgesic dose. This effect may be caused by a reduction in vascular uptake resulting from vasoconstriction or by a direct [small alpha, Greek]-agonist activity of epinephrine. 
Once in the CSF, fentanyl, similar to other opioids, spreads rostrally. The CSF concentration at the level of the cervical spine peaks within 20 min, compared with 3 h for morphine and 1 h for meperidine. Because of the high affinity of fentanyl with nonspecific binding sites in the lipid-rich spinal cord, only a small proportion (sup [tilde operator] 10%) of the administered dose migrates to the cervical region. 
Fentanyl also can migrate from the CSF into the epidural vascular compartment via the dura. However, little is known about the systemic pharmacokinetics of intrathecal fentanyl. At an average intrathecal infusion rate of 0.8 mg [middle dot] kg-1[middle dot] h-1after thoracotomy, vascular absorption is significant and fentanyl CPvalues range from 0.49-0.72 ng/ml. However, this range reflects CPvalues much lower than plasma analgesic concentrations when intravenous fentanyl infusion is used for post-thoracotomy analgesia (1.4-1.6 ng/ml). [17,18]This difference suggests that analgesic effect is mediated in part at the spinal level.
More than 40 published clinical trials document epidural fentanyl administration and effect. Most suggest that epidural fentanyl is less likely than morphine to produce clinically significant ventilatory depression. However, respiratory arrest has been reported. Fentanyl is reported to be associated with fewer minor adverse effects (nausea, vomiting, pruritus, sedation, urinary retention). [140,141]The modes of epidural administration are the same as those used intravenously, i.e., bolus dose alone, bolus dose combined with continuous infusion, continuous infusion alone, PCEA, and PCEA with a continuous fixed or variable background infusion. Both trauma and postsurgical populations have been studied, including patients with fractured ribs, those recovering from abdominal [19,127,143-147]and orthopedic surgery, [39,129]cesarean section, [38,148]thoracotomy, [17,18,52,130,131,149-153]and radical prostatectomy. In addition, epidural fentanyl often is combined with local anesthetic agents (particularly bupivacaine) and other adjuvant drugs (epinephrine, clonidine, other opioids) to provide postoperative analgesia, pain control for labor and delivery, and relief from perioperative cesarean section pain. These combined techniques are not discussed in this review.
Dose Requirements. The dosage of fentanyl used for repeated single bolus administration varies from 50 to 200 [micro sign]g (approximately 1-3 [micro sign]g/kg, [127,143,155,156]) and up to 5 [micro sign]g/kg. Analgesia begins within 15 min and lasts 2 to 4 h. [126,127,143,155,157]Increasing the diluent volume to 10-25 ml and using a concentration of 10 [micro sign]g/ml speeds onset time and increases the duration of action. However, the duration of analgesia provided by a bolus dose of fentanyl is brief, making administration by infusion necessary for adequate postoperative pain relief.
Epidural infusion rates range from 0.5-2.5 [micro sign]g [middle dot] kg-1[middle dot] h-1. Epidural or intravenous bolus opioid (usually fentanyl) supplementation may be used to achieve or maintain good analgesia, especially if the infusion rate is fixed. [52,144]In addition, PCEA has been used after operation, often with a background infusion, but in some cases alone. [19,67,147,159-161]Doses of 20-25 [micro sign]g with a lockout interval of 6-10 min and background infusion rates of 0.5-1 [micro sign]g [middle dot] kg-1[middle dot] h-1are most common. Variability in dose requirements, particularly in the first 24 h after operation, may be associated with differences in anesthetic technique (e.g., epidural vs. general anesthesia), intraoperative opioid administration, the magnitude of postoperative pain after different surgical procedures, or all of these. 
Analgesic Efficacy. Continuous epidural fentanyl infusion, PCEA, or both provide excellent analgesia and overcome the limitations to the duration of action associated with epidural bolus administration. Although analgesia is excellent at rest, pain scores increase with movement and coughing. [39,52,127,152,154,157]This is in contrast to the good analgesia at rest and with ambulation or coughing provided by intrathecal administration of fentanyl. [137,163]
Several investigators have compared the efficacy of intravenous and epidural fentanyl in a series of controlled trials conducted in different postoperative patient populations. [18,19,38,39,52,55,57,67,150]Whether the two techniques provide equivalently effective analgesia is controversial. To put the argument in perspective, several parameters are required for comparison: analgesic effect, dose requirement, CP, and respiratory and non-respiratory side effects. Equally important are the site and type of surgery and the site of epidural catheterization (i.e., lower abdominal or orthopedic surgery using a lumbar epidural catheter, [19,38,39,67]upper abdominal surgery or thoracotomy using a lumbar epidural catheter, [18,52,53,150]or upper abdominal surgery or thoracotomy using a thoracic epidural catheter [Table 7A]). [17,52,57]
Intravenous and epidural fentanyl provide equivalent analgesia in these surgical populations. Dose requirements are similar with the two modes of administration (e.g., a bolus plus infusion or PCEA). However, some investigators report that placing the catheter at a dermatomal level corresponding to the surgical site (thoracic placement for thoracotomy and upper abdominal surgery ) decreases the hourly and cumulative dose requirement for epidural delivery by 50% compared with intravenous. In contrast, other investigators report no significant differences in the analgesic dose requirements associated with thoracic versus lumbar epidural catheters for upper abdominal or thoracic surgery (Table 7B). [52,131,149,160]Similarly, neither the fentanyl CPassociated with analgesia nor the side effect profiles for the two techniques differ significantly. Overall, these findings indicate little clinical advantage to providing postoperative analgesia by the epidural rather than intravenous route of administration.
Side Effects. Most reviews of the nonrespiratory side effects of intrathecal and epidural opioids focus on morphine. [164-166]With epidural fentanyl, the most common side effect is pruritus, with an incidence of 0-85%. It usually appears within an hour after bolus epidural injection and lasts 20-30 min. Fentanyl-induced pruritus does not appear to be related to histamine release and can be antagonized with naloxone. Features that distinguish epidural fentanyl-induced pruritus from that resulting from epidural morphine include diminished intensity and localized (segmental or regional) effect. The facial area rarely is involved. [141,167]Cutaneous flare and urticaria are uncommon. [140,141]Treatment is rarely necessary.
The incidence of nausea and vomiting with epidural fentanyl is approximately 20-30%, which is comparable to other epidural opioids. Nausea usually occurs within 3 h of fentanyl administration. Treatment with antihistaminic agents, opioid agonist or antagonists, or pure opioid antagonists is common. [160,168]
Epidural opioids can produce dose-independent, naloxone-reversible urinary retention. The mechanism of this effect is believed to be secondary to the interaction of epidural opioids with opioid receptors in the sacral spinal cord. This interaction inhibits sacral parasympathetic neural outflow, relaxing the bladder detrusor muscle, increasing the maximum capacity of the bladder, thereby resulting in urinary retention. [170,171]The reported incidence of urinary retention ranges from 0-50% but is less than 12% in most studies. Data on the incidence of urinary retention after epidural fentanyl are limited by the common postoperative practice of placing or maintaining an indwelling urinary catheter.
Respiratory depression (Table 8) is the most serious adverse effect of the use of epidural opioids. Various factors may contribute to its occurrence with epidural fentanyl, including the use of additional opioids or sedative drugs, antiemetics, or prolonged infusions. [172,173]
Initially, the lipophilic properties of fentanyl were thought to confer minimal risk of delayed respiratory depression due to poor cephalad spread in the CSF. However, the single human CSF pharmacokinetic study of epidural fentanyl bolus administration at the lumbar level shows that maximum CSF concentrations in the cervical region are as much as 10% of CSF concentration in the lumbar region. 
Clinically significant respiratory depression can occur with either bolus doses or continuous infusions of epidural fentanyl. The carbon dioxide ventilatory response curve is depressed after an epidural bolus dose of 200 [micro sign]g. However, incidents of respiratory arrest [139,174]and profound respiratory depression [175-178]have been reported after a 100-[micro sign]g dose. Epidural infusions of fentanyl ranging from 0.5-1 [micro sign]g [middle dot] kg-1[middle dot] h-1depress the carbon dioxide ventilatory response curve. [129,179]Despite these data, clinically significant respiratory depression appears to be relatively rare with epidural fentanyl. Of those 29 studies with a total of approximately 600 patients that evaluated the incidence and severity of respiratory side effects, the overall incidence of clinically significant depression is approximately 1.8%. All these incidences of respiratory depression occurred in two of the studies. [131,150]In one study, administration of fentanyl through a thoracic or lumbar catheter at an average infusion rate of 1.55-2.06 [micro sign]g [middle dot] kg-1[middle dot] h-1resulted in respiratory depression requiring naloxone intervention in 4 of 30 patients. In the other study, it is difficult to assess the clinical significance of the respiratory depression because the protocol demanded arterial blood gas sampling every 2 h and administration of naloxone to each patient in whom the partial pressure of carbon dioxide in arterial blood exceeded 50 mmHg (7 of 25 patients), regardless of clinical status.
Epidural administration of fentanyl can provide good to excellent postoperative analgesia. Bolus administration produces a rapid onset but short duration of effect. A continuous infusion or PCEA is therefore more common in the postoperative setting and provides excellent analgesia at rest. (As with other routes of administration, the analgesic effect diminishes with movement and coughing.) Compared with intravenous administration, the analgesic efficacy, dose requirement, Cp, respiratory and nonrespiratory side effects of epidural and intravenous fentanyl are similar, indicating no advantage to using epidural rather than intravenous fentanyl infusion alone for postoperative analgesia.
Intrathecal fentanyl usually is combined with local anesthetic agents for perioperative anesthesia and analgesia, particularly in obstetric patients. [180,181]Only a limited number of reports, many of them single cases, document the use of intrathecal fentanyl alone for postoperative analgesia [182-184](Table 9). Modes of administration include single bolus injection, [183,185,186]intrathecal catheterization, repeated observer-administered or PCA boluses, [163,181]and continuous infusion via an intrathecal catheter. [137,184,187]Repetitive bolus injection of fentanyl via an intrathecal catheter also has been assessed as the sole source of analgesia during the first stage of labor. 
Dose Requirements. The minimum intrathecal bolus requirement for postoperative analgesia is 20 [micro sign]g. In obstetric patients, a smaller dose (10 [micro sign]g) is effective (Table 9). Onset of analgesia is usually within 5-15 min, but duration is variable, ranging from 1-5 h in most reports. Intrathecal fentanyl has also been administered as PCA via an intrathecal catheter. Although the ideal bolus size and lockout interval for intrathecal fentanyl via PCA remain unknown, good analgesia with no clinically significant adverse effects can be achieved with bolus doses of 6 [micro sign]g and a lockout interval of 1-1.5 h. 
At an average infusion rate of 0.8 [micro sign]g [middle dot] kg (-1)[middle dot] h-1, continuous intrathecal infusion produces satisfactory analgesia in patients having thoracotomy. A rate of 5 [micro sign]g/h (i.e., approximately 0.065 [micro sign]g [middle dot] kg-1[middle dot] h-1) in a similar surgical population resulted in inadequate post-operative analgesia and the need for parenteral opioid supplementation in nearly all patients. 
Analgesic Efficacy. Effective postoperative analgesia can be achieved with intrathecal bolus doses of 20 [micro sign]g (Table 9). Increasing this dose to 50 [micro sign]g produces excellent analgesia at rest, neutralizes the effects of ambulation or coughing on the quality of analgesia in patients after thoracotomy, and improves postoperative pulmonary function compared with PCA morphine alone. Similar analgesic efficacy also can be achieved with an intrathecal infusion at 0.8 [micro sign]g [middle dot] kg-1[middle dot] h-1, which also overcomes the painful effects of movement and coughing with greater speed than thoracic epidural or intravenous fentanyl at infusion rates of 1.22 and 1.27 [micro sign]g [middle dot] kg-1[middle dot] h-1, respectively. Thus, the intrathecal route provides better and more complete analgesia than fentanyl administered by other modalities that rarely provide dense analgesia with movement or coughing. A potential limitation to the use of the intrathecal technique is the need for an indwelling intrathecal catheter, which confers the risk for infection or neurotoxic effects (with local anesthetic agents), and necessitates clinical expertise in catheter placement and maintenance.
Side Effects. Side effects are relatively minor with intrathecal fentanyl, and only a few studies report any incidence of nausea or mild-to-moderate pruritus. However, these symptoms develop with both relatively large bolus doses (e.g., a 50 [micro sign]g bolus dose) and low-dose infusions (e.g., 5 [micro sign]g/h for 24 h;Table 9). A 30% incidence of urinary retention has been reported at a low-dose continuous infusion of 5 [micro sign]g/h progressively in 24 h. 
Clinically significant respiratory depression has not been reported with the use of intrathecal fentanyl (Table 9), even at total doses as high as 100 [micro sign]g delivered directly into the subarachnoid space. In contrast, the same dose of fentanyl (100 [micro sign]g) delivered epidurally has been associated with severe respiratory depression. Several explanations may account for this difference in effect. First, intrathecal administration results in lower systemic absorption than does epidural delivery. Second, the intrathecal dose requirement generally is lower than the epidural requirement, which further reduces the risk of dose-related ventilatory depression. Finally, studies of intrathecal fentanyl administration involve only small numbers of patients, potentially making it difficult to detect a statistically significant incidence of respiratory depression.
Compared with epidural administration, fentanyl delivery via the intrathecal route provides more intense and complete analgesia at rest and with movement, at a lower dose requirement than either the epidural or intravenous routes. Vascular absorption occurs but to a lesser extent. However, the requirement of an indwelling intrathecal catheter introduces the risk for infection or neurotoxic effects, limiting the popularity of this technique for use after operation.
Fentanyl is used widely as an analgesic agent in the postoperative or critically ill patient. Because of its physical properties and potency, it is effective via multiple routes of administration; noninvasive routes are being developed. Subarachnoid use provides the most intense, complete analgesia, although intravenous PCA, with its more convenient format, also is effective. Adverse effects are apparent with all modes of administration. Pruritus, urinary retention, and nausea and vomiting are common, and all patients receiving fentanyl for postoperative analgesia require vigilant monitoring to detect and treat respiratory effects. New experimental modalities, especially iontophoretic application and transmucosal delivery, present promising opportunities for postoperative analgesia.
The authors thank Christine Drane for preparing the manuscript, Winifred von Ehrenberg for editorial assistance, and Dr. Hossam El-Beheiry for advice to the manuscript.
[double dagger] Howell ST, Minto CF, Schlugman D, Glass PSA: Respiratory pharmacodynamics of bolus fentanyl in healthy volunteers (Abstract). Anesthesiology 1996; 85:A339.
[Section] Streisand JB, Busch MA, Gaylord BL, Gay MA, East KA: Dose proportionality of oral transmucosal fentanyl citrate in human volunteers (Abstract). Anesthesiology 1996; 85:A322.
[double vertical bar] Dsida R, Wheeler M, Birmingham P, Henthorn T, Avram M, Klein C, Cote C: A kinetic comparison: intravenous vs oral transmucosal fentanyl in tonsillectomy patients (Abstract). Anesthesiology 1996; 85:A1073