The true benefits of pain control infusion pump use are at least partially rooted in the placebo effect, and its use in TKA unpredictably contributes to postoperative pain management.
Few topics in orthopedic surgery have garnered as much interest as improvements in pain management. Pain control is essential to optimize patient comfort, assist with early postoperative physical therapy, avoid complications associated with prolonged bed rest, and reduce hospital inpatient durations and costs. In addition to the ethical responsibility of minimizing suffering, the Joint Commission has mandated that patients self-reported pain score must be documented as a fifth vital sign. All indications point to pain control ratings being used in the future not only as a measure of patient satisfaction, but as a potential measure of physician competence.
Surgeons have the insight, capability, and responsibility to play an active role in their patients postoperative pain control. Traditional uses of anti-inflammatory and narcotic medications serve as the backbone of most postoperative pain control regimens.1,2 Newer pain-relieving modalities, however, may offer additional advantages when used concurrently with these traditional methods.
Pain control infusion pumps automatically and continuously deliver local anesthetic through an intraoperatively placed catheter. Continuous local anesthetic models have been examined in the shoulder and knee.3,4 However, to date, no investigation has prospectively compared the potential analgesic effect of pain control infusion pump application in a system that allows for simultaneous direct comparison to a control model. Although evidence supports the use of pain control infusion pumps in cardiovascular, cardiothoracic, obstetric, and general surgical procedures, its role in orthopedic surgery is still being defined.
Within the field of orthopedics, joint replacement is a commonly and widely performed operation. Accordingly, joint arthroplasty has become a popular subject of recent research related to acute postoperative pain control in orthopedic surgery. We hypothesize that the use of pain control infusion pumps will significantly improve postoperative pain control in comparison to knees not receiving this intervention.
Materials and Methods
The present study was approved by our institutional review board (protocol number 2009-41). Inclusion criteria included all patients deemed appropriate for bilateral total knee arthroplasty (TKA). No exclusion criterion was used.
From June 2008 to June 2010, the knees of patients having bilateral TKA were randomized. The pain control infusion pump selected for use in our investigation was the ON-Q PainBuster pump (I-FLO Corporation, Lake Forest, California). Individuals with medical record numbers randomly generated ending in an even number had their right knee operated on first, with pain control infusion pump placement in the left. This was reversed with individuals with medical record numbers ending in odd numbers. We placed the pump in the second operated knee in an attempt to minimize accidental pump pullout during operating time.
Standard protocol used epidural anesthesia; however, general anesthesia was required per anesthesiologist discretion in a minority of cases as a consequence of patient comorbidity and/or preference. Total knee arthroplasty was performed through an anterior midline incision. Per individual surgeon discretion, arthrotomy was completed through lateral-parapatellar, midvastus, or subvastus incision. Infrequently, lateral release was performed and/or intra-articular suction drainage was included. Although these surgeon-specific technical deviations were present, these differences remained consistent between individual patient operative knees. Accordingly, each knee served as an exact control for the other knee.
A 0.5% bupivicaine solution was infused at 5 cc/hour for all surgeries. This was a deviation from normal technique for bilateral TKA. Typically, two 0.25% bupivicaine solutions were infused at 5 cc/hour to avoid cardiotoxity present with the larger infused volumes required for bilateral pump placement. Once patients completed the bilateral TKA, prior to closure, the intra-articular pump was placed according to standard industry-described technique (Figure 1) in the following fashion: The pain control infusion pump was primed with a 5 mL to verify catheter patency. After the protective guard was removed, the introducer was inserted through the skin approximately 3 to 5 cm from the surgical site. After the trocar was removed and withdrawn from the sheath, the catheter was advanced until the entire infusion catheter was visualized to be in the desired intra-articular location. Prior to final suturing, free movement of the catheter was verified. The catheter was primed a second time to ensure catheter patency. The catheter was coiled and secured with a Tegaderm dressing (3M, St Paul, Minnesota). A separate sterile dressing covered with a compressive wrap was placed on the coiled catheter and surgical site.
| || |
|Figure 1: Intra-articular pain control infusion pump catheter placement. Figure 2: Blinded and unblinded pain control infusion pump placement. |
The sham tubing was placed after wound closure was complete. Patient blinding was performed by 2 mechanisms: (1) the pain control infusion pump bulb and black bag was fastened to the patient on their midline (as opposed to the effected side) in an effort to conceal which knee contained the true intra-articular pain control infusion pump, and (2) an accessory tube (a standard component to all ON-Q Painbuster Pumps) was taped to the pain control infusion pump bulb. This additional tube exited the pain control infusion pump black bag and entered the knee bandage of the contralateral side. This sham tubing was secured distally with surgical tape. Superficially, wound bandages where applied identically, with no obvious indications revealing the sham tubing to the patient (Figure 2). The patient was then brought to the post-anesthesia care unit and transferred to the orthopedic floor in standard postoperative fashion.
Once on the orthopedic floor, twice-daily physical and occupational therapy commenced as scheduled. All patients where started on the following pain regimen: oxycodone and acetaminophen 5/325 mg, 1 to 2 tablets orally every 4 hours as needed for pain; celecoxib 200 mg orally twice a day; and morphine 4 mg intravenously every 2 hours as needed for breakthrough pain.
Additional narcotics were made available to the patients. Narcotic rescue was defined as the need to administer additional narcotics, including oxycodone or hydromorphone.
Vital signs were checked hourly for the first 4 hours, and then routinely checked every 4 hours until patient discharge from the hospital. Beginning at 4 hours after the patient arrived on the floor (after the hourly vital sign checks were finished), the nursing staff recorded normal vital sign information, including bilateral knee pain scores. Specifically, a pain score was recorded for each knee. This information was recorded, along with patient name, date, and time on a specific nursing sheet that was readily accessible to the nursing staff and was located in well-identified study notebook located in the patients room. Routine vital recording was continued until the pain control infusion pump was removed on postoperative day 3 during initial dressing change.
Just prior to this removal, the patient was asked by the investigators to discern any perceivable difference in postoperative pain between their knees. Once this response was recorded, the pain control infusion pump was removed, and its volume was recorded. Patient care then continued to follow normal standard protocol.
From June 2008 to June 2010, twenty-eight patients were identified at 2 institutions as candidates for bilateral TKA. Of this population, 24 patients completed the 3-day protocol. Upon completion of informed consent, 1 patient declined entry into the investigation. One patient had the pain control infusion pump accidentally removed in the post-anesthesia care unit. The 2 remaining patients were removed from the study after accidental premature dressing changes on the second postoperative day that unblinded the patient to the pain control randomization.
Five orthopedic surgeons (M.Z., B.E., D.R.) at 2 institutions performed the bilateral TKAs. One surgeon routinely performed the subvastus approach (66% of cases), 1 routinely performed a midvastus approach (8% of cases), and the remaining 3 performed a standard medial parapatellar approach (24% of cases). Patient demographics, as well as preoperative pain levels, knee deformity, hemovac drain placement, drain placement, lateral release, pain cocktail injection, anesthesia type, ability to identify the functional pain control infusion pump, and need for narcotic rescue, are outlined in Table 1.
Fourteen patients (58%) correctly perceived and correctly identified their knee with the functional pain control infusion pump. Narcotic rescue and the ability to correctly perceive and correctly identify knees with the functional pain control infusion pump are outlined in Table 2. Notably, of the 15 patients with uneven preoperative pain, 7 (47%) had the pump placed in the more painful knee. Of this group, 3 patients (43%) correctly perceived and correctly identified their knee with the functional pain control infusion pump. Of the 15 patients with uneven preoperative knee pain levels, 8 (53%) had the pain control infusion pump placed in the less painful knee. Of this group, 6 patients (75%) correctly perceived and correctly identified their knee with the functional pain control infusion pump, and 3 (38%) required postoperative narcotic rescue. Of the 5 patients who had general anesthesia, 4 patients (80%) correctly perceived and correctly identified their knee with the functional pain control infusion pump. Of the 19 patients who underwent epidural anesthesia, 10 (53%) correctly perceived and correctly identified their knee with the functional pain control infusion pump.
A total of 11 patients (46%) had their pain control infusion pump bulb volume measured on the removal of the pump on postoperative day 3 (Table 3). This component of the investigation was developed and incorporated midway into our research and is accordingly reflected with our inability to measure pump volume in all patients studied. Average volume was 168 mL (35% of original bulb volume) (standard deviation, 83.5; range, 70-370 mL). Of the 8 patients who correctly perceived and correctly identified their knee with the functional pain control infusion pump, average remaining bulb volume was 156 mL (33% of bulb volume remaining). Of the 3 patients who incorrectly identified their knee with the functional pain control infusion pump, average remaining bulb volume was 200 mL (42% of bulb volume remaining).
Effective postoperative pain management is a quintessential component of patient satisfaction. Pain control is imperative to ensure patient comfort, assist with postoperative physical therapy, avoid complications associated with prolonged bed rest, and reduce hospital inpatient durations and hospital costs. Pain control infusion pumps optimize postoperative pain control by delivering continuous flow of local anesthetic directly into the surgical site. To date, no investigation has prospectively compared the analgesic effect of pain control infusion pumps in a system that allows for simultaneous direct comparison to a control model.
Retrospective studies have documented success using continuous local anesthetic postoperative pain regimens. Pain reduction has been evidenced with cardiovascular,5 general,6 urologic,7 bariatric,8 plastic,9 and obstetric10 surgical cases. Successful orthopedic application has been found with anterior cruciate ligament reconstruction,3 rotator cuff repair,4 spinal fusion,11 and iliac graft harvest.12 Unfortunately, the chondrotoxity associated with local anesthetic delivery13-15 has made intra-articular application a dangerous and unpopular choice for postoperative pain control.16
Through removal of degenerated articular cartilage surfaces, joint arthroplasty would seem to be an ideal candidate to receive the analgesic relief of pain control infusion pumps, but avoid the chondrotoxicity associated with continuous local anesthetic delivery. Few studies have addressed this suspicion. Nechleba et al17 investigated the use of 0.25% bupivicaine intra-articular infusions in patients undergoing TKA. No significant improvement in pain relief or narcotic use was found between experimental and control groups receiving normal saline infusions. Mauerhan et al18 divided >100 patients undergoing TKA into groups receiving isolated intra-articular infusions of (1) normal saline, (2) morphine sulfate, (3) bupivicaine, or (4) a combination of morphine sulfate and bupivicaine. The total amount of postoperative pain medication used in the first 24 hours postoperatively was not statistically significant among the 4 treatment groups.
Similar results may be extrapolated from other arthroplasty surgical cases. Reeves and Skinner19 evaluated 66 knees undergoing unilateral TKA. Patients were randomized to receive normal saline, low-dose (0.2% ropivicaine) or high-dose (0.375% ropivicaine) local anesthetics. After 48 hours of continuous infusion, no benefit of local anesthetic could be identified. Chen et al20 evaluated 92 patients undergoing total hip arthroplasty. Patients were randomized to receive intra-articular infusions of normal saline or 0.5% bupivicaine. Although patients with continuous local anesthetic delivery demonstrated longer times to first narcotic use rescue, no difference was found in terms of amount of narcotic use, incidence of adverse events, hospitalization days, and Western Ontario and McMaster Universities Arthritis Index scores.
Our results failed to identify pain control infusion pumps as a reliable or predictable pain-relieving postoperative adjunct. Approximately 58% of patients evaluated were able to correctly perceive and correctly identify their knee with the functional pain control infusion pumps. When preoperative pain was equivalent, the pump seemed to offer little additional advantage. Patients with preoperative pain level disparities, however, demonstrated larger differences in the ability to correctly perceive and identify their knee with the functional pain control infusion pumps. Of this population, patients with the pain control infusion pump placed in the less painful knee were 32% more likely to correctly perceive and identify their knee with the functional pain control infusion pump. This suggests that pain control infusion pump function may be optimized when placed in the more painful side.
Individuals who underwent lateral release were approximately 20% more likely to correctly perceive and correctly identify their knee with the functional pain control infusion pump; however, conclusions are limited because of the small population size (4 patients). Patients who received general anesthesia were 27% more likely to correctly perceive and correctly identify their knee with the functional pain control infusion pump when compared to patients who received epidural placement. This suggests epidural use may obviate the analgesic effect of pain control infusion pump use. Benefit of pain control infusion pump application was not observed when evaluated between knee deformity type, pain cocktail injection, or hemovac drain use.
Narcotic rescue was required in 42% of patients studied. This group appeared to have little correlation with the ability to correctly perceive or identify their knee with the functional pain control infusion pump. Individuals with higher pain levels (requiring narcotic rescue) were 14% less likely to correctly perceive or identify their knee with the functional pain control infusion pump. This suggests patients with higher overall pain levels may benefit less from pain control infusion pump use. Despite our best efforts, pain scores were recorded intermittently. The subjectivity of assigned numerical values to identify pain levels was patient specific; however, we attempted to identify when, regardless of value, numerical equivalence was achieved between the 2 knees. Our goal was to potentially identify set periods with optimal pain relief using pain control infusion pumps within the window of the 3 days studies. Unfortunately, these inconsistencies made data interpretation to draw meaningful conclusions difficult. Otherwise, narcotic rescue did not appear to be dependent on preoperative pain discrepancy, preoperative knee deformity, anesthesia type, hemovac use, or pain cocktail administration.
Estimated volume based on a 3-day pump drainage rate of 5 mL/hour was 115 mL. On average, pumps delivered local anesthetic more slowly, at 4.3 mL/hour (86% of stated drainage rate). However, observed drainage rates ranged from 1.46 mL/hour to 5.56 mL/hour. This variability could be explained by pump injury during application, wound closure, or possibly during normal knee motion. Catheter compression could decrease flow, while catheter tearing could lead to increased flow. Surprisingly, patients who correctly perceived and correctly identified their knee with the functional pain control infusion pump tended to have less functional pumps (slower drainage rates of 1.41 mL/hour).
The placebo effect is a well-documented component of symptom treatment trials and may lead to the generation of false negative and false positive indices.21 The frequency of placebo effects in pain analgesic studies ranges from 1% to 60%.21 Factors identified that may serve to increase the magnitude of placebo response include patients expectations, clinicians warmth, prestige, positive attitude, and invasiveness of the intervention.22 Price23 was able to achieve near-complete pain relief for 19 hours in individuals with complex regional pain syndrome who received saline injection into stellate ganglion. It is difficult to assess the magnitude of our study; however, we believe some pain relief was attributable to the concept of the pain control infusion pump and not its actual pharmacologic effect. We believe this is evidenced in the variability of volume remaining in the pain control infusion pumps measured in patients who were able to correctly identify knees with functional pain control infusion pumps in place.
Interestingly, standard local anesthetic catheter placement technique is not well defined. The specific placement of the catheter may have important implications. Furthermore, the benefits of local anesthetic may be optimized in extra-articular locations in the soft tissues or near the incision.24 As some of the catheters are invariably accidentally prematurely partially pulled out, a minority of patients are likely receive the benefits of pain control infusion pump use as extra-articular pain relief. Intuitively, when placed on suction, effective drainage may work against pain control infusion pump infusion. It is also unclear how much of the local anesthetic remains in the surgical site. High concentrations of local anesthetic have been found in the copious drainage that nearly always occurs at the site of the catheterskin interface.17 This fluid may elute out some of the very analgesic ingredient that is being infused into the joint. Lastly, the ideal concentration of local anesthetic catheter has not been identified. The cardiotoxic effects of local anesthetic are real and dangerous. For these reasons, bilateral TKA patients receive twice the volume but half the local anesthetic concentration (0.25% bupivicaine) in each operative knee in comparison to patients undergoing unilateral procedures. This lower concentration may be inadequate to control pain. The interplay of these many factors certainly contributes to the effectiveness of continuous local anesthetic catheter ability to predictably control postoperative pain.
The use of a pain control infusion pump is not free from complication. Although infection has never been directly linked to continuous local anesthetic catheter use,25 direct intra-articular connection is a theoretic bacterial conduit that cannot be ignored. The tenuous dermal fixation makes the catheter prone to partial or complete catheter pullout. Lastly, local anesthetic catheters contribute to additional perioperative costs.
Our study is not free from limitations. The basis of our investigation assumes that postoperative pain levels between bilaterally operated knees are identical. Anecdotally, despite the practice of identical surgical technique and application of identical postoperative regimen, this may not always be the case. The small number of knees studied limits the significance of our results. We were limited by the surgical scheduling of bilateral TKA cases conducted while the principal investigator (E.A.) was completing his residency. The rate of nearly 1 bilateral TKA case per month could not be increased, as this operative intervention was discussed, planned, and implemented with prospective patients in standard fashion. To increase the number of patients studied, 5 surgeons at 2 separate institutions volunteered to offer this protocol to their patients. Although 1 surgeon performed 75% of the cases (18/24), our study is underpowered to assess whether pain control infusion pump effectiveness correlated with specific surgeon.
Furthermore, our conclusions on narcotic consumption are based only on the need for rescue narcotics; however, this analysis fails to evaluate the frequency of which the normal narcotic regimen was used. Therefore, in patients not receiving narcotic rescue, there may still have been a spectrum of narcotic consumption that we were not able to identify and potentially attribute to pain control infusion pump use. Lastly, our conclusions are reserved exclusively for pain control infusion pump use. The application of local anesthetic via local nerve block may offer an effective means of managing postoperative pain.
Although these surgeons all used the same TKA system, inevitable differences in surgical technique were present. Of note, 2 surgeons used a pain cocktail (ketorolac 15 mg, 20 cc 0.75% bupivicaine, 2 mg morphine sulfate, 0.3 mg epinephrine, and normal saline to create 60 cc solution), which was injected in the capsule and surrounding musculature of both knees prior to pump insertion. Although these differences existed and potentially affected postoperative pain levels, all surgeon-specific differences remained constant for both knees of each individual. Therefore, both knees for each patient existed in identically systemic conditions. We believe this unique comparison validates our results generated, and patients were able to prospectively and continuously compare pain control for both of their knees.
To minimize postoperative pain, many devices, including pain control infusion pumps, have been incorporated into normal postoperative regimens in an effort to increase patient satisfaction and potentially positively affect patient outcome. Unfortunately, many of these devices have no prospective randomized evidence supporting or outlining their specific use. We believe or study offers the first prospective blinded randomized study to help better define the role of pain control infusion pump use in joint arthroplasty. We found pain control infusion pumps to offer unpredictable and unreliable postoperative pain control. The need for narcotic rescue appeared unrelated to pain control infusion pump functionality. Trends indicated pain control infusion pump use could be most beneficial in the more painful preoperative leg in individuals receiving general anesthesia; however, interpretation of pump output rates failed to consistently correlate medication delivery to pain control. We conclude that the true benefits of pain control infusion pump use are at least partially rooted in the placebo effect, but we acknowledge that patients who received perceived pain relief overwhelmingly attributed analgesia to the pain control infusion pump, and accordingly, its contribution to pain postoperative pain relief, regardless of etiology, should not be ignored.
- Purtill JJ. Nonsteroidal anti-inflammatory drugs: are they safe after joint arthroplasty for early pain? Instr Course Lect. 2007; (56):133-137.
- Flory DA, Fankhauser RA, McShane MA. Postoperative pain control in total joint arthroplasty: a prospective, randomized study of a fixed-dose, around-the-clock, oral regimen. Orthopedics. 2001; 24(3):243-246.
- Hoenecke HR Jr, Pulido PA, Morris BA, Fronek J. The efficacy of continuous bupivacaine infiltration following anterior cruciate ligament reconstruction. Arthroscopy. 2002; 18(8):854-858.
- Gottschalk A, Burmeister MA, Radtke P, et al. Continuous wound infiltration with ropivacaine reduces pain and analgesic requirement after shoulder surgery. Anesth Analg. 2003; 97(4):1086-1091.
- Dowling R, Thielmeier K, Ghaly A, Barber D, Boice T, Dine A. Improved pain control after cardiac surgery: results of a randomized, double-blind, clinical trial. J Thorac Cardiovasc Surg. 2003; 126(5):1271-1278.
- Sanchez B, Waxman K, Tatevossian R, Gamberdella M, Read B. Local anesthetic infusion pumps improve postoperative pain after inguinal hernia repair: a randomized trial. Am Surg. 2004; 70(11):1002-1006.
- Mistry S, Miles B, Lipshultz L. Use of continuous local anesthetic infusion pumps in urological surgery. Contemporary Urology. 2004; (5):12-13.
- Cottam DR, Fisher B, Atkinson J, et al. A randomized trial of bupivacaine pain pumps to eliminate the need for patient controlled analgesia pumps in primary laparoscopic Roux-en-Y gastric bypass. Obes Surg. 2007; 17(5):595-600.
- Heller L, Kowalski AM, Wei C, Butler CE. Prospective, randomized, double-blind trial anesthetic infusion and intravenous narcotic patient-controlled anesthesia pump for pain management after free TRAM flap breast reconstruction. Past Reconstr Surg. 2008; 122(4):1010-1018.
- Gupta A, Perniola A, Axelsson K, Thörn SE, Crafoord K, Rawal N. Postoperative pain after abdominal hysterectomy: a double-blind comparison between placebo and local anesthetic infused intraperitoneally. Anesth Analg. 2004; 99(4):1173-1179.
- Bianconi M, Ferraro L, Ricci R, et al. The pharmacokinetics and efficacy of ropivacaine continuous wound instillation after spine fusion surgery. Anesth Analg. 2004; 98(1):166-172.
- Singh K, Samartzis D, Strom J, et al. A prospective, randomized, double-blind study evaluating the efficacy of postoperative continuous local anesthetic infusion at the iliac crest bone graft site after spinal arthrodesis. Spine (Phila Pa 1976). 2005; 30(22):2477-2483.
- Chu CR, Coyle CH, Chu CT, et al. In vivo effects of single intra-articular injection of 0.5% bupivacaine on articular cartilage. J Bone Joint Surg Am. 2010; 92(3):599-608.
- Piper SL, Kim HT. Comparison of ropivacaine and bupivacaine toxicity in human articular chondrocytes. J Bone Joint Surg Am. 2008; 90(5):986-991.
- Dragoo JL, Korotkova T, Kanwar R, Wood B. The effect of local anesthetics administered via pain pump on chondrocyte viability. Am J Sports Med. 2008; 36(8):1484-1488.
- Busfield BT, Romero DM. Pain pump use after shoulder arthroscopy as a cause of glenohumeral chondrolysis. Arthroscopy. 2009; 25(6):647-652.
- Nechleba J, Rogers V, Cortina G, Cooney T. Continuous intra-articular infusion of bupivacaine for postoperative pain following total knee arthroplasty. J Knee Surg. 2005; 18(3):197-202.
- Mauerhan DR, Campbell M, Miller JS, Mokris JG, Gregory A, Kiebzak GM. Intra-articular morphine and/or bupivacaine in the management of pain after total knee arthroplasty. J Arthroplasty. 1997; 12(5):546-552.
- Reeves M, Skinner MW. Continuous intra-articular infusion of ropivacaine after unilateral total knee arthroplasty. Anaesth Intensive Care. 2009; 37(6):918-922.
- Chen DW, Hsieh PH, Huang KC, Hu CC, Chang YH, Lee MS. Continuous intra-articular infusion of bupivacaine for post-operative pain relief after total hip arthroplasty: a randomized, placebo-controlled, double-blind study [published online ahead of print September 24, 2009]. Eur J Pain. 2010; 14(5):529-534.
- Turner JA, Deyo RA, Loeser JD, Von Korff M, Fordyce WE. The importance of placebo effects in pain treatment and research. JAMA. 1994; 271(20):1609-1614.
- Moore RA, Gavaghan D, Tramèr MR, Collins SL, McQuay HJ. Size is everythinglarge amounts of information are needed to overcome random effects in estimating direction and magnitude of treatment effects. Pain. 1998; 78(3):209-216.
- Price DD. Psychological and Neural Mechanisms of Pain. New York, NY: Raven Press; 1988.
- Andersen LØ, Husted H, Kristensen BB, Otte KS, Gaarn-Larsen L, Kehlet H. Analgesic efficacy of subcutaneous local anaesthetic wound infiltration in bilateral knee arthroplasty: a randomised, placebo-controlled, double-blind trial [published online ahead of print January 6, 2010]. Acta Anaesthesiol Scand. 2010; 54(5):543-548.
- Hovsepian RV, Smith MM, Markarian MK, et al. Infection risk from the use of continuous local-anesthetic infusion pain pumps in aesthetic and reconstructive abdominal procedures. Ann Plast Surg. 2009; 62(3):237-239.
Drs Argintar, Zawadsky, and Evans and Mr Armstrong are from Georgetown University Hospital, Washington, District of Columbia; and Dr Romness is from Virginia Hospital Center, Arlington, Virginia.
Drs Argintar, Zawadsky, and Evans and Mr Armstrong have no relevant financial relationships to disclose. Dr Romness is a paid consultant for TissueGene, Inc.
Correspondence should be addressed to: Evan Argintar, MD, Georgetown University Hospital, 3629 38th St NW #304, Washington, DC 20016 (email@example.com).