## Abstract

Background: Postoperative pain following augmentation mammaplasty can cause significant disability. In the authors' previously published prospective study of 644 consecutive augmentation mammaplasty patients, it was shown that the use of indwelling catheters for the postoperative instillation of bupivacaine is both safe and effective in postoperative pain management.

Objective: This study analyzes a large population of augmentation mammaplasty patients to not only compare the effectiveness of catheter control with systemic analgesics but also to delineate sites of pain and determine if any of a variety of factors influence postoperative pain.

Methods: Questionnaires were given to all patients to record the extent of pain reduction following the instillation of local anesthesia, patient preference for a variety of analgesics, and the locations of postoperative pain. Potential correlations to pain—including the number of children, age, handedness, preoperative fear, pain tolerance, and length of narcotic use—were investigated using statistical analysis. Operative aspects, such as the size of implants, use of asymmetrical implants, intraoperative expansion, duration of surgery, and sharp versus blunt dissection were also studied to determine if these factors influenced postoperative pain.

Results: Self-administration of bupivacaine the evening after surgery achieved pain reduction in 89% of patients. Patients gave high preference scores for the instillation of local anesthesia, comparable to rates achieved with the systemic narcotics Vicodin (Abbott Laboratories, Abbott Park, IL) and Percocet (Endo Pharmaceuticals, Chadds Ford, PA). Lower patient ratings were given to ibuprofen and methocarbamol. Patients identified multiple sites of postoperative pain, including the sternum, armpits, outer ribs, top of the breast (infraclavicular), interscapular region, and shoulders. Older patients experienced less pain. Patients who reported more pain, higher levels of fear, and sternal pain used narcotics for a longer period of time. Duration of surgery, blunt versus sharp dissection, and intraoperative expansion were not significantly related to pain. Combined implant size did not correlate with pain, though in the case of asymmetrical implants, greater pain (not statistically significant) on the larger side was reported.

Conclusions: Postoperative pain following augmentation mammaplasty may be influenced by several factors, but these appear unrelated to implant size, specific operative techniques (blunt versus sharp dissection), or duration of surgery. Pain may involve not only the breasts, but also the sternum, sides of the chest, armpits, and infraclavicular and interscapular areas. The use of indwelling catheters for the instillation of a long-acting anesthetic is rated comparable in efficacy to the systemic narcotics Vicodin and Percocet. Sternal pain can be severe and may require narcotics for effective pain control.

Medical personnel now consider the assessment of pain as the fifth vital sign. Significant physiologic consequences of inadequate pain treatment can include postoperative nausea and vomiting (PONV); increased heart rate, blood pressure, and myocardial contractibility; respiratory and abdominal muscle spasm; hypoxemia leading to pulmonary infection; reduced intestinal motility; urinary retention; increased platelet aggregation and thrombosis; impaired immune system; muscular weakness and fatigue; and fear, depression, and anxiety.1 Traditional systemic methods of pain control are not without their potential side effects, including altered mental status, narcotic dependence, nausea, constipation, depressed respiratory drive, and emesis.

Our previous studies determined that indwelling catheters were safe and effective as a conduit for the instillation of a long-acting local anesthetic, and that bolus administration of bupivacaine was comparable to continuous flow devices.2–4 In the present study, subgroups of more than 200 patients were analyzed to better understand postoperative pain after augmentation mammaplasty. Our investigations included the following:

• The response of pain to local anesthetic and analgesics, and patient preferences for various pain control methods;

• Assessment of implant size, intraoperative implant expansion, and the length of surgery in an attempt to identify any correlations between these parameters and the amount of postoperative pain experienced by the patient. Asymmetrical implants were evaluated to determine if more pain occurred on the side receiving the larger implant;

• Patient age, number of children, handedness, the patient's assessment of her own pain tolerance, and preoperative apprehension were studied to determine if there was any correlation to pain;

• The location of pain was assessed to address the needs of the patient more precisely;

• A small group of patients was randomized to compare continuous flow to bolus self-administration of bupivacaine.5 Another group of patients was randomized to compare pain following blunt versus sharp dissection of the subpectoral pocket; and

• The entire cohort was analyzed to count the number of days of catheter use, the incidence of complications, and revisions.

## Methods

In the more than 200 patients subjected to data analysis, all of the surgery was performed in a subpectoral plane using a breast fold incision under general anesthesia with bupivacaine local supplement. Since the beginning of the study, and continuing currently, all patients received a preoperative subcutaneous injection of 0.25% bupivacaine with epinephrine (0.25% Sensorcaine with 1:400,000 epinephrine [Astra-Zeneca, Wilmington, DE]), 20 mL per side, laterally, centrally, medially, and along the inframammary fold, as previously described.2 The inframammary fold was lowered as needed.6

The study included patients receiving both symmetrical and asymmetrical implants, a variety of constricted breast deformities, and augmentation mammaplasty revisions. Also included were patients undergoing various concomitant mastopexies. The study excluded breast reconstruction and breast reduction cases. Ancillary procedures were rarely performed at the same time. All patients were consecutive, and only prospective data were used. We learned early in the study that postoperative patients have short memories regarding the details of their discomfort.

Because two patients (nos. 324 and 343) developed hepatitis thought to be caused by cephalosporin sensitivity,7 prophylactic preoperative antibiotics were discontinued beginning with patient no. 349.

Patients and their caregivers were instructed both verbally and in writing on the use of the various analgesics and the proper hygiene for the catheter systems, such as the use of betadine ointment around the catheters, betadine scrub around the catheters when showering, and hand washing before touching the catheters.

Pain was recorded on each side using a numeric pain intensity scale of 1 to 10 (1 = no pain, 5 = moderate pain, and 10 = severe pain) both before instillation of local anesthesia and about 30 minutes after.

Before catheter removal (always performed by a nurse 30–60 min after instillation), the pocket was aspirated, often yielding blood-tinged fluid (Figure 1). This was especially helpful in a patient with von Willebrand disease to monitor the amount of postoperative bleeding.8 At times, the catheter also functioned as a drain.

Figure 1

Bloody aspirate is visible about 45 minutes after office instillation of 20 mL of bupivacaine into each catheter. The extension tubing is placed between the syringe and the catheter.

Figure 1

Bloody aspirate is visible about 45 minutes after office instillation of 20 mL of bupivacaine into each catheter. The extension tubing is placed between the syringe and the catheter.

### Pain Medication Ratings

We asked the patients to rate the various analgesics and local anesthesia on a scale of 1 to 5 (5 = best; 4 = very good; 3 = satisfactory; 2 = poor; and 1 = no help). Patients were asked to use only one analgesic at a time or the local anesthesia instillation. Because patients appeared to respond differently to analgesics on different days, only data collected in the first 24 hours were used.

Muscle relaxants (methocarbamol; Robaxin [Schwarz Pharma, Monheim, Germany]) were started 2 days preoperatively and continued four times a day for approximately 3 to 5 days postoperatively.

The day the patient was narcotic-free was recorded. A single dose counted as a “narcotic day,” and the patient was not considered to be “narcotic-free” until no narcotics were used.

### Pain Location

Surprisingly, we evaluated 207 patients before we thought of asking the patient, “Where is the pain?” One particular patient complained that “an elephant was sitting on my chest” and, on further questioning, she reported that the breasts were pain-free, but that she had severe pain in her sternum. At this time, the questionnaire was adjusted to record the various locations of any reported pain.

### Predictive Factors for Pain

Statistical tests were used to identify any correlations between the amount of pain experienced by the patient and factors as follows: age; childbirth; the patient's assessment of their own pain tolerance (scored 1–5; 1 = can handle pain easily; 2 = pretty good about pain; 3 = average; 4 = not so great with pain; 5 = I'm a real baby and hate pain); handedness (right or left); preoperative apprehension (scored 1–10; 1 = no apprehension; 10 = maximum apprehension); the size of implants (for symmetric implants: “Did larger implants cause more pain?”; for asymmetrical implants: “Did the side with the larger implant cause more pain?”); intraoperative expansion; duration of surgery; and blunt versus sharp dissection.

Twenty-four–hour pain scores were used to compare blunt versus sharp dissection in 22 patients. Consecutive patients were randomized using sharp dissection and blunt dissection on opposite sides using dual plane augmentation mammaplasty. All patients underwent sharp dissection of the inferior pole of the breast using the technique of Tebbetts.9 Sharp dissection continued superiorly on one side and blunt dissection using a urethral sound was performed on the opposite side.

### Statistical Analysis

Data collection and statistical analysis was performed using Excel (Microsoft, Redmond, WA), and JMP (SAS Institute, Cary, NC). Because this was not a double-blind prospective study designed to test a limited number of relationships, data analysis was treated as exploratory. To identify potential interesting relationships meriting further study, no correction for multiple hypothesis testing was performed. Where possible, all relationships were tested for statistical significance using both parametric and nonparametric tests. For all reported correlations, the results of these two approaches were in concordance.

## Results

### Distribution of Patients Across Treatment Regimens

Of the 687 patients who received indwelling catheters for the control of postoperative pain, 437 patients were given two doses of bupivacaine to self-administer at home; 188 patients in the early portion of the study underwent administration by the nursing staff only; 41 patients had a continuous flow catheter on one side and a self-administration catheter on the other side (comparison study5); and 21 patients had only continuous flow catheters.

### Duration of Catheter Devices

Of 687 patients, catheters were maintained for 1 day postoperatively by 583 patients (85%) and for 2 days by 104 patients (15%).

A previous paper from our study analyzed the safety and efficacy of bolus administration. It was found that the instillation of bupivacaine was as effective as Vicodin (Abbott Laboratories, Abbott Park, IL) with ibuprofen and Percocet (Endo Pharmaceuticals, Chadds Ford, PA), and that about 90% of patients experienced pain relief following instillation of bupivacaine.2

### Response to Analgesics and Local Anesthesia

Figure 2 shows no statistical difference between the following analgesics and local anesthesia: Vicodin 5 mg/500 mg (Abbott Laboratories) plus 600 mg of ibuprofen (mean pain score rating, 4.15; n = 79), self-administration of bupivacaine (4.13; n = 243), Percocet 5 mg/325 mg (Endo Pharmaceuticals) alone (3.96; n = 89), On-Q continuous pump (3.95; n = 41), and Vicodin alone (3.93; n = 149). Ibuprofen (3.28; n = 190) and methocarbamol (3.27; n = 242) were statistically inferior. Methocarbamol was found to be more effective in patients with interscapular pain than patients without interscapular pain (n = 185; t test, P = .0196; Wilcoxon rank-sum, P = .0231). Patients with interscapular pain rated the narcotics Vicodin and Percocet worse than patients without interscapular pain (Percocet: n = 89; t test, P = .0201; Wilcoxon rank-sum; P = .0345; Vicodin: n = 149; t test, P = .0039; Wilcoxon rank-sum, P = .0102). With the exception of these differences, none of the other pain locations responded differentially to the analgesics tested.

Figure 2

Analgesic preference. Box plots above the name of each medication represent the range of ratings of effectiveness of each analgesic. Analgesics are presented in order of mean response from most effective (best = 5), to least effective (no help = 1). While the first five treatments were consistently rated effective, ibuprofen and Robaxin received statistically lower scores. The colored diamond is centered on the mean response and ends at the 95% confidence intervals.

Figure 2

Analgesic preference. Box plots above the name of each medication represent the range of ratings of effectiveness of each analgesic. Analgesics are presented in order of mean response from most effective (best = 5), to least effective (no help = 1). While the first five treatments were consistently rated effective, ibuprofen and Robaxin received statistically lower scores. The colored diamond is centered on the mean response and ends at the 95% confidence intervals.

### Day Narcotic-Free

As might be expected, the day narcotic-free correlates with 24-hour mean pain scores (correlation, 0.37; n = 127; P < .0000; Figure 3, A). While patients with sternal pain did not consistently rate narcotics higher than patients without sternal pain, these patients used narcotics for a longer period of time (mean day narcotic-free with sternal pain, 3.53 days; without, 2.17 days; N = 129; t test, P = .0013; Wilcoxon rank-sum, P = .0015). This is consistent with the clinical observation that patients with severe sternal pain seem to respond better to narcotics than other analgesics or the instillation of local anesthesia.

Figure 3

Pre- and postoperative correlates of pain. A sampling of the relationship between average pain scores recorded in the first 24 hours postoperatively and possible predictors of the degree of discomfort experienced by the patient. Each panel indicates the mean 24-hour pain score on the y-axis with covariants on the x-axis. A, Relationship between 24-hour pain scores and the number of days the patient used narcotics. Patients reporting greater discomfort were more likely to use narcotics for longer periods of time. B, Relationship between preoperative apprehension and pain scores. Patients reporting greater apprehension were also more likely to report greater discomfort. C, Relationship between patient age and pain scores. Older patients generally reported less discomfort. D, Relationship between the length of surgery and pain scores. There is no clear correlation between the length of the surgical procedure and pain experienced by the patient. In fact, the point circled in red indicates the single longest surgery in the study with pain scores well below average. Red lines indicate a best-fit linear regression through all points on the graph. The green line is a bivariate normal ellipse enclosing 90% of the data.

Figure 3

Pre- and postoperative correlates of pain. A sampling of the relationship between average pain scores recorded in the first 24 hours postoperatively and possible predictors of the degree of discomfort experienced by the patient. Each panel indicates the mean 24-hour pain score on the y-axis with covariants on the x-axis. A, Relationship between 24-hour pain scores and the number of days the patient used narcotics. Patients reporting greater discomfort were more likely to use narcotics for longer periods of time. B, Relationship between preoperative apprehension and pain scores. Patients reporting greater apprehension were also more likely to report greater discomfort. C, Relationship between patient age and pain scores. Older patients generally reported less discomfort. D, Relationship between the length of surgery and pain scores. There is no clear correlation between the length of the surgical procedure and pain experienced by the patient. In fact, the point circled in red indicates the single longest surgery in the study with pain scores well below average. Red lines indicate a best-fit linear regression through all points on the graph. The green line is a bivariate normal ellipse enclosing 90% of the data.

### Predictors of Pain

#### Preoperative apprehension

Patients reporting lower preoperative apprehension also reported lower 24-hour mean pain scores (Figure 3, B; correlation, 0.2272; P = .0031).

#### Childbirth and age

Patients with children reported lower mean 24-hour pain scores (5.03 vs 5.95; n with children = 118, n without children = 31; t test, P = .0150; Wilcoxon rank-sum; P = .0174; Figure 3, C). Older patients were more likely to have children, and older patients also reported less pain during the first 24 hours. However, in both patients with and without children, a significant decrease in the pain reported was observed with the increasing age of the patient, suggesting that older patients are likely to report less pain regardless of whether or not they have borne children.

#### Duration of surgery

No correlation was seen between length of surgery and the mean 24-hour pain scores reported by patients (Figure 3, D). This was true of both the total patient population (n = 232; correlation, −0.006; P = .9276) and patients receiving augmentations only (n = 186; correlation, −0.046; P = .5325). The red circle in Figure 3, D indicates a 2.5-hour augmentation-only procedure, with less than average pain.

#### Implant size and pain

Interestingly, when the entire patient cohort was analyzed, there was no correlation between the combined implant size and the amount of pain reported by the patient (n = 276; correlation, −0.02; P = .6996). However, patients with asymmetrical implants (n = 79; 29%) tended to have slightly more pain on the side with the larger implant regardless of handedness (n = 76; correlation, 0.25; paired t test, P = .0084; Wilcoxon rank-sum, P = .012). We also noted that in the 204 patients receiving symmetrical left and right implants, 128 (63%) had some asymmetry in their mean 24-hour pain scores, with a slight bias toward greater pain on the side of the stronger hand, but this difference did not reach statistical significance.

#### Blunt versus sharp dissection

In the comparative study of 22 patients, no significant difference was found in the pain scores (matched pairs, sharp vs blunt, n = 22; sharp mean 24-hour pain score = 4.20, blunt mean 24-hour pain score = 4.34; P = .7267).

#### Intraoperative expansion

Among 238 patients, 197 (83%) had intraoperative expansion ranging from 5 cc to 140 cc (mean, 41 cc). Forty-one patients had no expansion. Where expansion was used, 25 cc accounted for 37% of the cases and 50 cc for 28% of cases. Patients had neither more nor less pain with expansion.

#### Pain location

Figure 4 summarizes the location of pain experienced by the patient. In some instances, more than one site was painful, and both the intensity and location fluctuated during different time intervals. The most common sites of pain were the breast (n = 257/284; 90%) and the sternum (n = 216/284; 76%). Other sites included the armpits (n = 113/175; 65%), outer ribs (sides of the chest, n = 142/284; 50%), top of the breast (infraclavicular area; n = 127/284; 45%), interscapular region (n = 104/284; 37%), and the shoulders (n = 28/176; 16%).

Figure 4

Pain location. A summary of the frequency of pain experienced in each location expressed as a percentage of patients reporting pain in that location.

Figure 4

Pain location. A summary of the frequency of pain experienced in each location expressed as a percentage of patients reporting pain in that location.

“Morning stiffness” was not studied separately but was found to be a common complaint of patients after sleeping. It was found that implant mobility exercises were more helpful than analgesics in treating this problem.

### Pain Regression

Figure 5 displays the diminution of pain over 7 days.

Figure 5

Pain regression over 7 days: This figure shows the pooled pain scores collected over the full 7-day period. The colored diamond is centered on the mean response and ends at the 95% confidence intervals. (RR = recovery room; po0 = day of surgery; OV, office visit; po1 = day after surgery (1 day postoperative); po2 = 2 days postoperatively; po3 = 3 days postoperatively; po4 = 4 days postoperatively; po5 = 5 days postoperatively; po6 = 6 days postoperatively; po7 = 7 days postoperatively.).

Figure 5

Pain regression over 7 days: This figure shows the pooled pain scores collected over the full 7-day period. The colored diamond is centered on the mean response and ends at the 95% confidence intervals. (RR = recovery room; po0 = day of surgery; OV, office visit; po1 = day after surgery (1 day postoperative); po2 = 2 days postoperatively; po3 = 3 days postoperatively; po4 = 4 days postoperatively; po5 = 5 days postoperatively; po6 = 6 days postoperatively; po7 = 7 days postoperatively.).

### One-Month Pain Study

An attempt was made to determine the use of analgesics for an additional 3 weeks after the 7-day study was concluded (patient nos. 301 to 320). Eleven patients did not return the questionnaire. Of the 9 patients who did, 0 were taking narcotics, 3 used no medication, and 6 patients used ibuprofen or Tylenol for 8 to 28 days (average, 16 days). These patients complained of residual nipple soreness, a feeling of tightness in the chest, breast engorgement, difficulty sleeping, incision discomfort, unilateral dull pressure, and tenderness to touch.

## Complications and revisions

### Infection

Complications were no greater than would be expected without the catheters. One patient developed a superficial Staphylococcus aureus infection that first appeared on her right side 10 days postoperative and was treated with Duricef (Warner Chilcott, Inc., Rockaway, NJ). The infection cleared, but recurred three times on the right side during the next 8 months. This was treated each time with Duricef with subsequent clearing. Surprisingly, the infection reappeared on the opposite side (left breast) 2 weeks later (8 months postoperatively) with clearing once again. At 2 years and 2 months post surgery, the infection reappeared on her left side and responded to antibiotics. Two months later, her recurrent infection became an abscess with exposure of the left implant, requiring explantation of the left implant. The patient was examined by infectious disease specialists and to date we remain unsure of the precise etiology; possibilities include systemic infection related to the patient working in microbiology with both Staphylococcus and Streptococcus; work in a doctor's office; contamination at the time of surgery, or contamination of the catheter system. Cultures were negative for acid-fast bacilli.

### Hematoma

Three unilateral hematomas were evacuated the day after surgery (3/1374 breasts; 0.2%). None of these patients developed capsular contracture. Pain catheters were inserted after hematoma evacuation.

### Capsular Contracture

Capsular contractures with a Baker score of II to III developed in seven unilateral implants and one bilateral implant (9/1374 breasts; 0.7%).

### Revisions

Eight unilateral implants and four bilateral implants required lowering (12/1374; 0.9%). Three unilateral and five bilateral implants bottomed out (average implant size, 310 cc; range, 250 to 350 cc) requiring inferior capsulorrhaphy and raising of the implants (0.06%). Twelve patients requested exchange for larger implants and two patients for smaller implants (14/687; 2.0%). Ten patients underwent implant exchange to correct unilateral deflation (10/1374 breasts; 0.07%); no other cases were reported during this 11-year period.

### Other

Temporary finger numbness was reported by some patients. A kinked catheter could be corrected by advancing the catheter. On rare occasions, the catheter was accidentally pulled out by the patient.

## Discussion

In our 11 years of experience (December 1996 to December 2007) with 687 patients, we have found that the instillation of local anesthesia into the dissected pocket of an augmentation mammaplasty is effective and safe.2 No known overdoses of bupivacaine occurred, and the amount of local anesthesia used was within acceptable ranges. Both continuous flow and intermittent bolus self-administration systems (see Table) are effective in controlling pain. The advantage of continuous flow is that this system does not require patient intervention. The advantage of self-administration is that patients have control over the timing and need for self-administration. In self-administration, instillation is only done when the patient has discomfort. Self-administration is also considerably less expensive than continuous flow ($10 vs$200). The cost factor is the primary reason that most of our patients choose the intermittent flow catheters, though both devices were discussed and patients were given a choice between the two.

At the beginning of this study, we were concerned about an external catheter being in contact with an implant because of the potential for contamination leading to infection. In addition, we were concerned about patients self-administering their own local anesthesia. Despite these concerns, we encountered only one unusual infection of unknown etiology (leading to unilateral explantation) and had no problems with patients using their medications.

We have observed that postoperative pain after breast enlargement is highly variable, and that there is a wide range of both pain experienced and individual response to different analgesics. Even for any single patient, the response to a pain treatment may be more effective at certain times and less effective at others. Many patients fear postoperative pain, and we find that preoperative apprehension correlates with 24-hour pain scores and narcotic use. The considerable insight into postoperative pain gained from this study has increased our ability to have an honest discussion of the pain process during the consultation and to give the patient reassurance that we will do everything possible to keep them comfortable after their surgery. Patients are grateful for our concerns and for the added postoperative analgesia. Reduction of preoperative apprehension has become an important part of our pain management program.

We find that patients prefer Vicodin with ibuprofen, local anesthesia using catheters, and Percocet over ibuprofen alone or methocarbamol (Figure 3). Schneider10,11 found that 80% of subpectoral procedures responded well to the muscle relaxant methocarbamol. Hidalgo12 noted that while preoperative intercostal nerve blocks do not significantly decrease postoperative pain, a preoperative dose of methocarbamol might be of benefit. In our study, we found a wide variation of responses to methocarbamol. Though there may be additional benefit for the patient complaining of interscapular pain, the overall response to this drug appears to be poor.

Table.

Current Protocol for Self-Administration and Continuous Flow Catheters

 Self-Administration Twenty mL of 0.25% bupivacaine HCl 1:400,000 epinephrine each side Preoperative infiltration: 40 mL Before discharge: 40 mL Evening dose: 40 mL Next day: 40 mL Day-after office visit: additional 40 mL before implant mobility exercises On-Q System (I-Flow Corporation, Lake Forest, CA) Continuous flow 0.25% bupivacaine HCl without epinephrine Preoperative infiltration: 40: mL Two mL per hour per breast = 96 mL Recommended allowance Each dose = 225 mg (90 mL) Total 24-hour dose = 400 mg (160 mL)
 Self-Administration Twenty mL of 0.25% bupivacaine HCl 1:400,000 epinephrine each side Preoperative infiltration: 40 mL Before discharge: 40 mL Evening dose: 40 mL Next day: 40 mL Day-after office visit: additional 40 mL before implant mobility exercises On-Q System (I-Flow Corporation, Lake Forest, CA) Continuous flow 0.25% bupivacaine HCl without epinephrine Preoperative infiltration: 40: mL Two mL per hour per breast = 96 mL Recommended allowance Each dose = 225 mg (90 mL) Total 24-hour dose = 400 mg (160 mL)

To aid patients in achieving return to normal activities within 24 hours, Tebbetts13 recommends that they use only ibuprofen for analgesia. By contrast, we found that patients rate ibuprofen as significantly less effective than other pain control methods (Figure 2). The apparent contrast between these results may reflect the fact that Tebbetts' article was not a pain study but rather a determination of return to normal daily activities, and did not include an evaluation of other pain control methods. When patients are given more control over their own pain management choices, ibuprofen alone ranks poorly among the available pain control techniques.

A number of different sites are involved in the pain process following augmentation mammaplasty (Figure 4). Sternal pain is often quite disabling and appears to respond best to narcotics and local anesthesia instillation, whereas back pain (interscapular pain) is uncomfortable and responds well to methocarbamol. Knowing the location of a patient's pain is helpful in counseling the patient on how to use their medications.

There is no benefit to depriving patients of narcotic pain control if they experience severe pain that is unresponsive to nonnarcotic medication. Though some patients require guidance from the physician to avoid addiction to these drugs, patients generally know when narcotics are no longer needed. The ability of the patient to appropriately taper their own narcotic use is clearly demonstrated by the relationship between reported pain scores and day narcotic-free (Figure 3, A).

According to Tebbetts,13 a longer operation requires more medications, “contributing to a longer and more difficult recovery.” However, we find no correlation between the length of the surgical procedure and the pain reported by the patient or the duration of narcotic use (Figure 3, D and data not shown). It has always been the senior author's opinion that a carefully performed procedure, regardless of the time involved, is of paramount importance; this view is clearly supported by our data and by Mahabir et al.14

In the same study, Tebbetts13 states that blunt dissection of the pocket should be avoided. However, in our pilot study of 22 patients there was no significant difference between pain resulting from blunt or sharp dissection.

It may be intuitive to think that larger implants cause more pain. However, when the entire patient cohort was analyzed, no correlation was seen between implant size and pain, although very small nonstatistical differences were detected when asymmetrical implants were analyzed within a given patient. These patients reported slightly more pain on the side of the larger implant. There appears to be no correlation between the degree of intraoperative expansion and pain.

The etiology of pain following augmentation mammaplasty is for the most part unknown. Biggs15 postulates that trauma to the ribs, bleeding into the dissected pocket, and trauma to the breast tissue itself in the process of retraction may increase postoperative pain. Tebbetts13 feels that in addition to the duration of surgery, tissue trauma, bleeding, and inflammation are additional causative factors. It has been the senior author's opinion that postoperative pain is related to the sudden stretch of tissues combined with the trauma of surgical dissection. Our longest operation (2.5 hrs; Figure 3, D [red circle]) was caused by difficulty in controlling a retracted perforator in the inferomedial portion of the dissection. Considerable char resulted from repeated attempts at coagulating the bleeding vessel, yet this patient reported less than average pain.

It would be helpful to know ahead of time if there were certain subsets of patients who could be predicted to have more pain. Our results suggest that older patients and patients with less fear going into the procedure will report less pain (Figure 3, C). This is consistent with the observation that younger patients often seem to be “out of control” in the recovery room and that they complain of more severe pain during the postoperative period.

The retention of the catheters for a second day, as used in 15% of our patients, is beneficial for those patients with continued pain, can be useful for early weaning from narcotics, and is helpful for patients who are simply afraid of experiencing postoperative pain. Patients are offered additional days of catheter pain control; however, none of our patients wanted this pain control for more than 2 days.

## Literature review

The history of pain control using catheters to deliver local anesthesia dates back to at least 1935.16 In this German article, Capelle used intermittent doses of a local anesthetic delivered through subcutaneous, preperitoneal metal cannulae to help prevent postoperative pneumonitis. He reports a measurable improvement in diaphragmatic excursion, and therefore ventilation, often to preoperative parameters. This resulted in better pain control and respiratory excursions than the administration of morphine.

At a time when procaine was the local anesthetic of choice (and lasted for only 40 to 60 min), Ansbro,17 in a 1946 article, described continuous brachial plexus block using an indwelling blunt needle for upper extremity procedures.

The use of indwelling catheters for pain control, using local anesthesia instilled into a variety of surgical sites, is currently expanding. Most of the studies reviewed were statistically analyzed, controlled, blinded, and randomized, and demonstrate a clear trend confirming the efficacy of pain control using indwelling catheters with instillation of a long-acting local anesthetic.

Pain management and the use of pain pumps were discussed in a recent practice advisory.18 A metaanalysis of 44 randomized controlled trials (enrolling a total of 2141 patients) reported the efficacy of continuous wound catheters with improved analgesia, reduced opioid use, a reduction in side effects, such as PONV, increased patient satisfaction, and perhaps reduced hospital stay.19 These benefits were noted across a wide range of surgical procedures, with a low incidence of catheter-related complications.

Morrison et al20 reported on the value of pain pumps following mastectomy in conjunction with the use of suction drains. Baroody et al21 noted the efficacy of continuous pain pumps following immediate autologous breast reconstruction using a latissimus dorsi pedicled flap, demonstrating decreased use of oral and intravenous pain medications, less pain, and less PONV compared to the control group. Losken et al22 found that patient-controlled analgesia requirements were significantly lower in a study of consecutive transverse rectus abdominis myocutaneous (TRAM) flap breast reconstruction patients, compared to those patients who did not have the pumps. Heller et al23 used continuous flow catheters for 5 days following free TRAM flaps, noting reduced narcotic use. Mahabir et al15 noted that the combination of locally administered intraoperative ketorolac and bupivacaine with epinephrine “significantly reduced pain in the postoperative period.” Pain in the recovery room appeared to be reduced by the use of infusion pumps in breast reduction and breast reconstruction patients, according to a study by Lu and Fine.24 Paul25 has found pain pumps to make a remarkable difference in postoperative pain in both breast augmentation and abdominoplasty, but Bray et al26 did not find any benefit from their use in abdominoplasties.

Many of the articles reviewed were from the orthopedic literature and described the use of continuous flow catheters as an extension of preoperative nerve blocks. As noted in an editorial by Klein,27 this technique is associated with sustained, effective postoperative analgesia; opioid sparing; improved rehabilitation, and improved patient well being with minimal side effects. Klein28 studied 40 adult patients scheduled for open rotator cuff repair comparing a continuous flow interscalene brachial plexus block using 0.2% ropivacaine to saline in a prospective, randomized, double-blinded, placebo-controlled study. He demonstrated a significant reduction in morphine consumption (P = .0004) and significant reduction in pain (P < .001) in the continuous flow group.

In a controlled study of 30 patients undergoing moderately painful, upper extremity orthopedic surgery, Ilfeld et al29 noted that a continuous infraclavicular nerve block with ropivacaine significantly reduced pain, and that patients had less pain with movement, used fewer oral narcotics, and had overall greater satisfaction, without complications. Enneking et al30 also reported the benefits of continuous flow following upper extremity amputation for malignancy. Gottschalk,31 in a study of 45 patients, found that 3.75 mg/mL of ropivacaine as a continuous wound infiltration for 48 hours after shoulder surgery resulted in significantly reduced pain scores with less opioid requirements. Catheters have been left in the axillary plexus sheath for up to 16 days without untoward effects.32,33

Hoenecke,34 in two studies involving knee surgery, compared different flow rates of bupivacaine to controls. The 2-mL per hour flow rate showed similar results to 5-mL per hour, and both studies demonstrated reduced pain and reduced narcotic usage compared to controls. In the author's experience of 500 patients, there were no toxic effects of the local anesthetic, no infections related to the catheter, and no wound healing problems. White et al35 studied the use of a continuous popliteal sciatic nerve block after foot and ankle surgery and found that 0.25% bupivacaine at 5 mL per hour resulted in a statistically significant reduction in pain scores and opioid use, leading to improved patient satisfaction and quality of recovery compared to controls.

Although most of the literature reviewed deals with continuous flow, some authors feel that bolus anesthesia, as needed, is superior to continuous infusion, because not all patients have postoperative pain. Rawal et al,36 in a study of 60 patients scheduled for ambulatory hand surgery, compared ropivacaine to bupivacaine using catheters to maintain an axillary plexus blockade by intermittent bolus anesthesia. A reduction in pain scores was seen with both drugs. On the day of surgery, ropivacaine was noted to be more effective (P < .05); however, the differences were not statistically significant by the second postoperative day. In another report on intermittent bolus infusions, Gupta et al37 left a catheter in the bed of the gallbladder following laparoscopic cholecystectomy. In a controlled study of 40 patients (ropivacaine vs saline, n = 20 each) using bolus infusions of 10 mL of 0.5% ropivacaine as needed for postoperative pain, it was found that during the first 4 postoperative hours, the ropivacaine group had statistically significant lower scores for deep (visceral) pain, and pain during coughing. After 4 hours, pain in general was mild in intensity in both groups. Although there was no difference in pain, Baig et al,38 in a controlled, blinded elective colectomy study, noted earlier ambulation and significantly less narcotic requirements in patients with continuous catheter anesthesia for 72 hours.

Other specialties have reported on the success of postoperative catheters. Pain after cesarean delivery was studied by Givens et al39 and Fredman et al.40 Givens noted significant reduction of narcotic use in the bupivacaine group with 1 infection in this group. In Fredman's study of 50 patients, the use of “rescue” morphine was significantly lower (P < .01) and pain scores generated after coughing and leg raises were significantly lower (P < .04) in the ropivacaine group. Givens et al39 indicate that it is uncertain how much pain is attributable to the incision compared to uterine pain. It their opinion, the elimination of some of the superficial components of pain after cesarean delivery could modulate the perception of deeper visceral pain. The benefits of continuous flow are noted by Zimberg et al41 and Ng et al42 following elective abdominal hysterectomy. In another study of continuous delivery of 0.5% bupivacaine after abdominal hysterectomy, Leong et al43 found no significant differences in pain or morphine usage compared to controls.

Vintar et al,44 Lau et al,45 Sanchez et al,46 and LeBlanc et al47 report benefits of continuous flow local anesthesia flowing inguinal hernia repair, whereas Schurr et al48 found only modest improvements in pain scores. Similar benefits were noted in colon and rectal surgery by Faria49 and Kumar50; managing trochar site pain after operative laparoscopy by Stringer el al51; and following median sternotomy by White et al.52

In a review article, Liu et al53 summarized pertinent anatomy, technical aspects, and current evidence when available in prospective randomized trials for the indications and efficacy of continuous perineural techniques for postoperative analgesia. A variety of blocks were discussed in addition to the review of the literature.

The delivery rate accuracy of 6 portable infusion pumps was studied by Ilfeld et al.54 Infusion rate accuracy differed significantly among the pumps, exhibiting flow rates ± 15% of their expected rate for 18% to 100% of their infusion duration. An increase in temperature also affected pumps to differing degrees, with infusion rates increasing from 0% to 25% for each model tested.

Bergman et al55 reported on neurologic complications of 405 consecutive continuous axillary catheters in 368 patients. Nine complications were noted in eight patients (2.2%). These included localized infection, axillary hematoma, retained catheter fragment requiring surgical excision, and two patients with signs and symptoms of systemic (preseizure) local anesthetic toxicity. The incidence of neurological complications with continuous axillary blockade was similar to that of single-dose techniques.

Severe complications appear to be rare but have been described. Garg et al56 reported on a death associated with an indwelling orbital catheter in a patient undergoing revision of an orbital implant. At autopsy, the catheter was noted to have passed through the superior orbital fissure with evidence of the injection of marcaine into the subarachnoid space.

In a personal verbal communication, W. Edwards, Jr., MD, discussed the 10-year use of continuous epidural narcotic infusions in aortic surgery, which allows for a more rapid recovery and less postoperative pain. An isolated use of low-dose local anesthesia masked an aortic graft occlusion leading to a postoperative paralysis due to spinal cord ischemia.

There continues to be widespread belief in the efficacy of preemptive analgesia for the control of postoperative pain. A metaanalysis by Møiniche57 included only double-blind, randomized, comparisons of identical or nearly identical analgesic regimens initiated before versus after surgical incision. According to Møiniche, “The review revealed a lack of evidence for preemptive treatment with nonsteroidal antiinflammatory drugs, intravenous opioids, intravenous ketamine, peripheral local anesthetics, and caudal analgesia to be of any benefit with respect to postoperative pain relief compared with a similar postincisional treatment.”57

## Conclusion

An 11-year prospective study of 687 consecutive patients with augmentation mammaplasty shows that indwelling catheters for the postoperative self-administration instillation of bupivacaine is both safe and effective in the management of postoperative pain. This modality gives the patient an additional means for pain control that can be used in conjunction with systemic analgesics. Sites other than the breast are involved in the pain process, and sternal pain is especially debilitating.

Fear of postoperative pain appears to be an important predictor of the amount of postoperative pain. Older women appear to have less pain, and patients who have more pain, or a greater fear of postoperative pain, tend to use narcotics for a longer period of time. Intraoperative expansion, the duration of surgery, and sharp versus blunt dissection appear to be unrelated to postoperative pain. Combined implant size is also unrelated, although in the case of asymmetrical implants, greater pain (not statistically significant) on the larger side may be reported.

## Disclosures

The authors have no financial interest in and received no compensation from manufacturers of products mentioned in this article.

## Acknowledgments

The authors would like to acknowledge Gloria Smith for relentlessly pursuing the completed questionnaires and the expertise of Janet Lynn in database maintenance and manuscript preparation; Daniel A. Sterling, MD, for his valuable assistance; and Fred C. Storm, MD, and Wayne Stadelman, MD, for their translation of the Capelle article.16

## References

1.
Joshi
GP
Ogunnaike
BO
Consequences of inadequate postoperative pain relief and chronic persistent postoperative pain
Anesthesiol Clin North America

2005
;
23
:
21
26
.
2.
Pacik
PT
Nelson
CE
Werner
C
Pain control in augmentation mammaplasty: Safety and efficacy of indwelling catheters in 644 consecutive patients
Aesthetic Surg J

2008
;
28
:
279
284
.
3.
Pacik
PT
Werner
C
Jackson
N
Lobsitz
C
Pain control in augmentation mammaplasty: The use of indwelling catheters in 200 consecutive patients
Plast Reconstr Surg

2003
;
111
:
2090
2096
.
4.
Pacik
PT
Werner
C
Follow-up: Pain control in augmentation mammaplasty: The use of indwelling catheters in 350 consecutive patients
Plast Reconstr Surg

2005
;
115
:
575
577
.
5.
Pacik
PT
Pain management in augmentation mammaplasty: A randomized comparative study to compare the use of a continuous infusion versus self-administration intermittent bolus of a local anesthetic
Aesthetic Surg J

2004
;
24
:
523
530
.
6.
Pacik
PT
Augmentation mammaplasty: Enhancing inferomedial cleavage
Aesthetic Surg J

2005
;
25
:
359
364
.
7.
Pacik
PT
Augmentation mammaplasty: Postoperative cephalosporin induced hepatitis
Plast Reconstr Surg

2007
;
119
:
1136
1137
.
8.
Pacik
PT
Augmentation mammaplasty and von Willibrand's disease
Plast Reconstr Surg

2005
;
115
:
1441
1442
.
9.
Tebbetts
JB
Dual plane breast augmentation: Optimizing implant-soft tissue relationships in a wide range of breast types
Plast Reconstr Surg

2001
;
107
:
1255
1272
.
10.
Schneider
MS
Pain reduction in breast augmentation using methocarbamol
Aesthetic Plast Surg

1997
;
21
:
23
24
.
11.
Schneider
MS
Methocarbamol: Adjunct therapy for pain management in breast augmentation
Aesthetic Surg J

2002
;
22
:
380
381
.
12.
Hidalgo
DA
Pusic
AL
The role of methocarbamol and intercostal nerve blocks for pain management in breast augmentation
Aesthetic Surg J

2005
;
25
:
571
575
.
13.
Tebbetts
JB
Achieving a predictable 24-hour return to normal activities after breast augmentation: Part I. Refining practices by using motion and time study principles
Plast Reconstr Surg

2002
;
109
:
273
290
.
14.
Mahabir
RC
Peterson
BD
Williamson
JS
Valnicek
SM
Williamson
DG
East
WE
Locally administered ketorolac and bupivacaine for control of postoperative pain in breast augmentation patients
Plast Reconstr Surg

2004
;
114
:
1910
1916
.
15.
Biggs
TM
Discussion: Pain control in augmentation mammaplasty: The use of indwelling catheters in 200 consecutive patients
Plast Reconstr Surg

2003
;
111
:
2097
2098
.
16.
Capelle
W
Die Bedeutung Des Wundschmerzes und Seiner Ausschaltung Fur Den Ablauf Der Atmung Bei Laparotomierten
Dtsch Z Chir

1935
;
246
:
466
485
(in German).
17.
Ansbro
EP
A method of continuous brachial plexus block
Am J Surg

1946
;
71
:
716
722
.
18.
Iverson
RE
Lynch
DJ
Practice advisory on pain management and postoperative nausea and vomiting
Plast Reconstr Surg

2006
;
118
:
1060
1069
.
19.
Liu
SS
Richman
JM
Thirlby
RC
Wu
CL
Efficacy of continuous wound catheters delivering local anesthetic for postoperative analgesia: A quantitative and qualitative systematic review of randomized controlled trials
J Am Coll Surg

2006
;
203
:
914
932
.
20.
Morrison
JE
Jr
Jacobs
VR
Reduction or elimination of postoperative pain medication after mastectomy through use of a temporarily placed local anesthetic pump vs. control group
Zentralbl Gynakol

2003
;
125
:
17
22
.
21.
Baroody
M
Tameo
MN
Dabb
RW
Efficacy of the pain pump catheter in immediate autologous breast reconstruction
Plast Reconstr Surg

2004
;
114
:
895
898
.
22.
Losken
A
Parris
JJ
Douglas
TD
Codner
MA
Use of the infusion pain pump following transverse rectus abdominis muscle flap breast reconstruction
Ann Plast Surg

2005
;
54
:
479
482
.
23.
Heller
L
Kowalski
A
Butler
CE
Randomized, controlled, double-blind evaluation of local anesthesia infusion pumps for donor site pain management following free TRAM flap breast reconstruction
Plast Reconstr Surg

2006
;
118
:
69
suppl
24.
Lu
L
Fine
NA
The efficacy of continuous local anesthetic infiltration in breast surgery: Reduction mammaplasty and reconstruction
Plast Reconstr Surg

2005
;
115
:
1927
2034
.
25.
Paul
M
Breast augmentation and abdominoplasty: Postoperative management with pain pumps
Aesthetic Surg J

2005
;
25
:
69
71
.
26.
Bray
DA
Jr
Nguyen
J
Craig
J
Cohen
BE
Collins
DR
Jr
Efficacy of a local anesthetic pain pump in abdominoplasty
Plast Reconstr Surg

2007
;
119
:
1054
1059
.
27.
Klein
SM
Beyond the hospital: Continuous peripheral nerve blocks at home
Anesthesiology

2002
;
96
:
1283
1285
.
28.
Klein
SM
Grant
SA
Greengrass
RA
Nielsen
KC
Speer
KC
White
W
et al
Interscalene brachial plexus block with a continuous catheter insertion system and a disposable infusion pump
Anesth Analg

2000
;
91
:
1473
1478
.
29.
Ilfeld
BM
Morey
TE
Enneking
FK
Continuous infraclavicular brachial plexus block for postoperative pain control at home. A randomized, double-blinded, placebo-controlled study
Anesthesiology

2002
;
96
:
1297
1304
.
30.
Enneking
FK
Scarborough
MT
EA
Local anesthetic infusion through nerve sheath catheters for analgesia following upper extremity amputation. Clinical report
Reg Anesth

1997
;
22
:
351
356
.
31.
Gottschalk
A
Burmeister
MA
P
Krieg
M
F
Kreissl
S
et al
Continuous wound infiltration with ropivacaine reduces pain and analgesic requirement after shoulder surgery
Anesth Analg

2003
;
97
:
1086
1091
.
32.
Ang
ET
Lassale
B
Goldfarb
G
Continuous axillary brachial plexus block–A clinical and anatomical study
Anesth Analg

1984
;
63
:
680
684
.
33.
Sarma
VJ
Long-term continuous axillary plexus blockade using 0.25% bupivacaine. A study of three cases
Acta Anesthesiol Scand

1990
;
34
:
511
513
.
34.
Hoenecke
HR
Continuous local anesthetic infiltration
Orthopedic Technology Review

2002
;
3
:
1
4
.
35.
White
PF
Issioui
T
Skrivanek
GD
Early
JS
Wakefield
C
The use of a continuous popliteal sciatic nerve block after surgery involving the foot and ankle: Does it improve the quality of recovery?
Anesth Analg

2003
;
97
:
1303
1309
.
36.
Rawal
N
Ailvin
R
Axelsson
K
Hallen
J
Ekback
G
Ohlsson
T
et al
Patient controlled regional anesthesia (PCRA) at home. Controlled comparison between bupivacaine and ropivacaine brachial plexus analgesia
Anesthesiology

2002
;
96
:
1290
1296
.
37.
Gupta
A
Thorn
SE
Axelsson
K
LG
Agren
G
Holmstrom
B
et al
Postoperative pain relief using intermittent injections of 0.5% ropivacaine through a catheter after laparoscopic cholecystectomy
Anesth Analg

2002
;
95
:
450
456
.
38.
Baig
MK
Zmora
O
Derdemezi
J
Weiss
EG
Nogueras
JJ
Wexner
SD
Use of the ON-Q pain management system is associated with decreased postoperative analgesic requirement: Double blind randomized placebo study
J Am Coll Surg

2006
;
202
:
297
305
.
39.
Givens
VA
Lipscomb
GH
Meyer
NL
A randomized trial of postoperative wound irrigation with local anesthetic for pain after cesarean delivery
Am J Obstet Gynecol

2002
;
186
:
1188
1191
.
40.
Fredman
B
Shapiro
A
Zohar
E
Feldman
E
Shorer
S
Rawal
N
et al
The analgesic efficacy of patient-controlled ropivacaine instillation after cesarean delivery
Anesth Analg

2000
;
91
:
1436
1440
.
41.
Zimberg
SE
Davila
G
Seiler
J
Neimark
M
On-Q pain management system in elective abdominal hysterectomy
Am J Obstet Gynecol

2002
;
99
:
S35
(abstr).
42.
Ng
A
Swami
A
Smith
G
Davidson
AC
Emembolu
J
The analgesic effects of intraperitoneal and incisional bupivacaine with epinephrine after total abdominal hysterectomy
Anesth Analg

2002
;
95
:
158
162
.
43.
Leong
WM
Lo
WK
Chiu
JW
Analgesic efficacy of continuous delivery of bupivacaine by an elastomeric balloon infusor after abdominal hysterectomy: A prospective randomized controlled trial
Aust N Z J Obstet Gynaecol

2002
;
42
:
515
518
.
44.
Vintar
N
Pozlep
G
Rawal
N
Godec
M
Rakovec
S
Incisional self-administration of bupivacaine or ropivacaine provides effective analgesia after inguinal hernia repair
Can J Anesth

2002
;
49
:
481
486
.
45.
Lau
H
Patil
NG
Lee
F
Randomized clinical trial of postoperative subfascial infusion with bupivacaine following ambulatory open mesh repair of inguinal hernia
Dig Surg

2003
;
20
:
285
289
.
46.
Sanchez
B
Waxman
K
Tatevossian
R
Gamberdella
M
B
Local anesthetic infusion pumps improve postoperative pain after inguinal hernia repair: A randomized trial
Am Surg

2004
;
70
:
1002
1006
.
47.
LeBlanc
KA
Bellanger
D
Rhynes
VK
Hausmann
M
Evaluation of continuous infusion of 0.5% bupivacaine by elastomeric pump for postoperative pain management after open inguinal hernia repair
J Am Coll Surg

2005
;
200
:
198
202
.
48.
Schurr
MJ
Gordon
DB
Pellino
TA
Scanlon
TA
Continuous local anesthetic infusion for pain management after outpatient inguinal herniorrhaphy
Surgery

2004
;
136
:
761
769
.
49.
Faria
TA
.
The On-Q pain management system in the control of postoperative pain in colon and rectal surgery
.
Presented during the 9th United European Gastroenterology Week
,
Amsterdam, The Netherlands
,
October 6–11, 2001
(abstr).
50.
Kumar
S
.
Continuous postoperative analgesia in anorectal surgery–Hemorrhordectomy
.
Presented at the American Society of Colorectal Surgeons Annual Meeting
,
San Diego, CA
,
June 3–8, 2002
(abstr).
51.
Stringer
NH
Rodino
KL
Edwards
M
Kumari
N
On-Q system for managing trochar site pain after operative laparoscopy
J Am Assoc Gynecol Laparosc

2000
;
7
:
552
555
.
52.
White
PF
Rawal
S
Latham
P
Markowitz
S
Issioui
T
Chi
L
et al
Use of a continuous local anesthetic infusion for pain management after median sternotomy
Anesthesiology

2003
;
99
:
918
923
.
53.
Liu
SS
Salinas
FV
Continuous plexus and peripheral nerve blocks for postoperative analgesia
Anesth Analg

2003
;
96
:
263
273
.
54.
Ilfeld
BM
Morey
TE
Enneking
FK
The delivery rate accuracy of portable infusion pumps used for continuous regional analgesia
Anesth Analg

2002
;
95
:
1331
1336
.
55.
Bergman
BD
Hebl
JR
Kent
J
Horlocker
TT
Neurologic complications of 40 consecutive continuous axillary catheters
Anesth Analg

2003
;
96
:
247
252
.
56.
Garg
S
Piva
A
Sanchez
RN
AA
Death associated with an indwelling orbital catheter
Ophthal Plast Reconstr Surg

2003
;
19
:
398
400
.
57.
Møiniche
S
Kehlet
H
Dahl
JB
A qualitative and quantitative systematic review of preemptive analgesia for postoperative pain relief: The role of timing of analgesia
Anesthesiology

2002
;
96
:
725
741
.