Europace Advance Access originally published online on May 5, 2007
Europace 2007 9(7):545-550; doi:10.1093/europace/eum060
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
IMPLANTABLE CARDIOVERTER DEFIBRILLATOR
Can we implant cardioverter defibrillator under minimal sedation?
1 Department of Cardiology A, Hôpital Cardiologique de Lille, CHRU, Blvd du Pr Leclercq, F-59037 Lille (Cedex), France; 2 Department of Cardiology C, Hôpital Cardiologique de Lille, CHRU, Lille, France; 3 Department of Cardiology, Roubaix, France; 4 Department of Anaesthesia, Hôpital Cardiologique de Lille, CHRU, Lille, France
Manuscript submitted 29 March 2006. Accepted after revision 8 March 2007.
* Corresponding author. Tel: +33 3 20 44 50 38; fax: +33 3 20 44 68 98. E-mail address: c-marquie{at}chru-lille.fr
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Aim In a prospective study, we tested the feasibility of implantable cardioverter-defibrillator (ICD) implantation under local anaesthesia (LA) with minimal sedation (MS) vs. short general anaesthesia (SGA) for defibrillation test (DT).
Methods and results We implanted ICDs in 118 patients between October 2002 and November 2003. Surgery was performed under LA with MS. Depending on the day of admission, patients had DT without SGA with a shock delivered when patient is unconscious (MS group, n = 73) or with short general anaesthesia (SGA group, n = 45). The patients were asked to rate the intensity of pain on a 10-point visual analogue scale (VAS) at the end of the implantation procedure and for the patient of MS group just after DT (VAS-DT). Visual analogue scale was not influenced by the type of anaesthesia (MS vs. SGA). Univariate analysis showed that NYHA >III, 3 leads implanted and an incomplete understanding of the explanation on the procedure were risk factors of VAS >4. Only understanding of the explanation on the procedure has an influence on pain in multivariate analysis.
Conclusion ICD implantation under MS even for DT is feasible and acceptable for the patient.
Key Words: Implantable cardioverter-defibrillator (ICD), Anaesthesia, Pain, ICD implantation method
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Recent studies indicate that placement of an implantable cardioverter defibrillator (ICD) for sudden death's primary prevention1
,2 in patients with LV systolic dysfunction significantly reduces overall mortality at intermediate-term follow-up. Thus, the rates of device implantation will probably increase greatly worldwide. The implant procedure has been simplified with improvements in technology. The ICD could be implanted practically like a pacemaker under local anaesthesia (LA)3–7 with minimal sedation (MS). However, general anaesthesia or deep sedation is always required for the assessment of defibrillation efficacy during implantation. This requirement can limit the growth of ICD placement in some centres and increases the cost of ICD placement. The goal of this single-centre study was to compare prospectively the feasibility of ICD implantation under LA with MS vs. SGA for defibrillation test (DT). |
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
In our centre, an ICD implantation can be scheduled with SGA only on Monday morning and Wednesday afternoon: SGA group. Apart from these two half days, implantations were performed under LA with MS even during defibrillation testing: MS group. Depending on their day of hospital admission, patients were included in the MS or SGA groups. Patients were informed for the nature of anaesthesia (MS or SGA). If a patient expresses a particular wish for MS or SGA, he was excluded of this study. All patients signed a written informed consent and the study was approved by our institutional research committee.
Exclusion of the study
Exclusion criteria were
- Medical necessity: sub pectoral implant, requirement of general anaesthesia during all the procedure (children).
- Expression of a particular wish for LA or SGA by patient.
Population
We implanted transvenous lead ICDs in 122 patients between October 2002 and November 2003.
Implantation procedure
In all patients, surgery was performed under LA with lidocaine hydrochloride (10 mg/mL) and MS with sublingual midazolam (5 mg/mL, 1 mg/10 kg if age 
70-years-old or 0.5 mg/10 kg if age > 70-years-old) and intravenous nalbuphine (5 mg if age 75-years-old). All the implantation procedures were made in the same operative room. An infraclavicular incision was made and a pocket was made in the prepectoral subfascial space. Transvenous electrodes were introduced into the venous system by left cephalic vein cut down or transcutaneous cannulation of the left subclavian vein if the cephalic access was unsuccessful. Leads were inserted under fluoroscopic guidance. An infraclavicular incision was made. A pocket was fashioned in the prepectoral subfascial space. The device was anchored with sutures to the pectoral muscle and deep fascia. The skin was closed in three layers.
Explanation on the different steps of the implantation's procedure
The information on the implant procedure was given to the patient within the preceding 24 ho by the implanting physician. The explanations on the DT were again given before the ventricular fibrillation (VF) induction. The same blinded observer performed all the interviews, and the patient was asked to re-explain the procedure at his entry in the operating room and to re-explain the DT immediately before the VF induction to be sure that he understood explanations and to classify the patient in two groups before DT: complete understanding of the procedure or incomplete understanding of the procedure.
Defibrillation test
The defibrillation was tested by a single shock at 20 J at the end of the ICD implantation for all patients.
In the MS group, only MS, performed before implantation, was used for defibrillation testing: no general anaesthesia. In the SGA group, SGA was used just before defibrillation testing.
In the MS group, no anaesthesiologist was present in the operating room. Defibrillation testing was made in the manual mode; we induced VF, we waited for that patient to become unconscious and a shock was delivered. If a 10 J safety margin was not obtained after two VF inductions or if VF was non-inducible with direct or alternating 30, 40, or 50 Hz, an anaesthesiologist was called in the operating room and the subsequent tests were performed in the automatic mode with a SGA.
In the SGA group, before induction of VF and defibrillation threshold testing, a SGA was achieved by an anaesthesiologist with thiopental or propofol. Oxygen saturation was monitored continuously and oxygen was supplied by a face mask. The DTs were done in the automatic mode.
Whatever the group was, the choice of VF induction mode (direct or alternating 30, 40, or 50 Hz) depended on the ICD manufacturers. The current for VF induction was delivered during 4 then 6 and finally 8 s. No T-wave shocks were used without SGA. Implantable cardioverter-defibrillators used and the VF induction modes were Marquis (n = 34; VR, DR, and Insync) from Medtronic with induction by 50 Hz delivered through a dedicate bipole; Atlas (VR, DR) and Epic (HF, DR) from St Jude (St Jude Medical, USA) (n = 27) with induction by direct current; Alto 2 DR from ELA medical (n = 8) with induction by 30 Hz; Ventak Prizm (n = 21) with induction by 40 Hz delivered through a bipole including the electrodes of defibrillation (coils and can) and Vitality (VR; n = 22) with induction by 40 Hz delivered through an integrate bipole and Belos (n = 6; VR, DR) from Biotronik (biotronic GmbH, Germany) with induction by 40 Hz delivered through an integrate bipole. All the implanted ICDs were active can devices.
Pain evaluation
The same blinded observer performed all the interviews. Patients were asked to rate pain intensity on a 10-point visual analogue scale (VAS) from 0 (no pain) to 10 (worst imaginable pain) at the end of implantation procedure or when he was capable of answering. We used the visual analogue thermometer, which is an adapted version of the VAS that does not require any written intervention by the patient. This scale is commonly used to evaluate pain during invasive procedure or surgery made under local anaesthesiology or MS.8–10 It consists of a rigid plasticized cardboard strip on white colour with a horizontal black opening 10-cm long by 2-cm wide (Figure 1). The left and right extremities of this opening are identified by the expressions ‘no pain’ and ‘unbearable pain’. The opening is covered with a black opaque band that slides from left to right by means a trip located on the back of the scale. On the back, there is also a 10-cm ruler graduated to the nearest millimetre with the extremities corresponding to the exact demarcation limits. As the strip is moved across the opening, the increasing intensity of pain is shown by the black band. The more intense the pain, the more the black band lengthens toward the limit of ‘unbearable’. The observer repositions the strip and then slowly moves the strip toward the right, asking the patient to stop at the point corresponding of the pain to be described. The numerical value associated with this point can be read to the nearest millimetre from the back of the device.11
|
We have chosen the cut-off of 4 to separate procedure with tolerable pain (moderate pain) and procedure with high pain.12
In the MS group, we evaluated pain just after DT with the VAS just to evaluate pain due to the DT: induction and shock [visual analogic scale after defibrillation test (VAS-DT)].
Statistical analysis
Statistical analysis was performed with SPSS software, version 11.5 (SPSS Inc. Chicago, IL). Mean values ± SD were calculated for quantitative data. The quantitative variables were compared between groups with use of unpaired Student's t-tests. Qualitative variables were compared with use of the Pearson's 
2 test or the Fisher's exact test when necessary. Logistic regression analysis was used to determine variables that were independently associated. For non-parametric values we used Wilcoxon test. P < 0.05 was considered significant.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Population
Of 118 patients (mean aged 61.4 ± 14 years), 122 were included. 73 and 45 in the MS and SGA groups, respectively. Four patients were excluded of the study: two patients have asked an implantation under SGA and the two remaining patients has refused SGA. Patient's characteristics for the two groups are summarized in Table 1. A majority of the implanted defibrillation leads were single coil leads (60% in both groups). In both groups 60 and 40% of leads were implanted by cephalic or subclavian approach, respectively. 64.5% of patients were implanted with a double chamber ICD and 7% avec a biventricular ICD. The repartition of VVI, DDD, and biventricular ICD was equivalent in the two groups: MS or SGA group.
|
The total procedure duration was 84 ± 35 and 86 ± 40 min in the MS and SGA groups, respectively (ns). The blood pressure, heart rates, and oxygen saturation were comparable in the both groups before incision, before VF induction, after VF induction, and at the end of the procedure. No patient required fluid volume loading or drugs to maintain haemodynamic parameters during the procedures.
Induction methods
The induction mode tested, the number of delivered shock, and VF duration in both groups are summarized in Table 1. A total of 121 and 73 VF inductions were attempted in the MS and the SGA groups, respectively (P = ns).
Direct current was more effective than alternating current to induce VF with 35/38 (92.1%) and 78/111 (70.3%) successful VF inductions with direct current and alternating current, respectively (P = 0.007). In the SGA group, T-wave shock was effective to induce VF in 76,2%.
The initial mode of induction (alternating) failed in only two patients in the MS group, requiring SGA for T-wave shock induction. For nine patients in the SGA group, T-wave shock was ineffective to induce VF and direct or alternating current was required. The first 20 J defibrillation shock were uneffective in 11 (15%) and 1 (2.2%) patients in the MS and the SGA groups, respectively (P = 0.03). Therefore, the number of shock is higher in the MS group (Table 2).
|
The duration of VF before shock is higher in the MS group than in SGA group (Table 2).
Pain
The type of anaesthesia (MS/SGA) did not influence VAS (Figure 2).
|
Univariate analysis (Table 3) did not find any effect of the type of anaesthesia on the pain. The understandings of the procedure, NYHA classification, and number of leads have an influence on pain. In this analysis, induction's modes tend to influence the results of VAS: direct current is more painful than alternative current.
|
If we analysed pain during DT (VAS-DT), in the MS group, we find that DC is more painful than AC with 50% of patients with a VAS-DT > 4 for 15% of patients inducing by AC (P = 0.0006) (Table 4). The configuration of alternating current during VF induction influenced VAS-DT with 53, 85, and 85% patients with VAS-DT = 0 with alternating current through electrodes of defibrillation, integrate bipole, and dedicate bipole, respectively (P = 0.02). Univariate analysis (Table 4) found that NYHA
3, 3 leads implanted, induction mode, shock felt, and an incomplete understanding of the explanation of the procedure steps were risk factors of pain (VAS-DT > 0) or high intensity pain (VAS-DT > 4) during DT.
|
Multivariate analysis (Table 5) showed that only understanding of procedure has an influence of pain. For MS procedure the number of lead and the mode of induction influence pain level during DT.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Our study showed that DT could be performed without SGA and that the type of anaesthesia did not influence the pain's level at the end of an ICD implant procedure. Evaluation of ICD's ability to restore sinus rhythm is required during a defibrillator placement procedure. This evaluation was performed under SGA. With the increasing number of ICDs implanted, this practice must be justified. The first question is: is it really necessary to test defibrillation threshold at the time of implantation? The response is probably ‘yes’ because an inadequate safety margin may occur in >6% of patients.13
At this time, it seems still necessary to test defibrillation threshold even in primary prevention indications because one cause of high defibrillation threshold is low ejection fraction.13 Concerning the length of implantation, there is no difference between the two groups because SGA is not time consuming.
These results show that the mode of VF induction has an influence on pain. We have some data in the literature about sedation for atrial or external defibrillation, but no literature about sedation and ventricular defibrillation.14,15 Direct current and alternating current delivered between the can and the right ventricular defibrillation coil are more painful than alternating current. In these two painful induction modes, the current induces muscular contractions which may be painful. Alternating current through an intra-cardiac bipole (dedicate or integrated) does not induce muscular contractions and is painless. In this study, we did not use T-wave shock under MS. In our experience this induction mode had been so painful and poorly tolerated that we have decided to never perform T-wave shock on a conscious patient. There are no data in other publications about pain induced by the different induction methods.
From the Table 3, we can say that 24% (11/45) of patients from SGA group had no pain during procedure for 20% (15/73) from MS group (P = ns). In the same time, we can extrapolate that 18% (8/45) of patients from SGA group vs. and 27% (20/73) from MS group have a VAS score 4 (P = ns). Even if there is no difference statically some patients in MS group could have a procedure painfully. In this situation, we think that a good understanding of the different steps of the implant procedure decreased the pain perceived by the patient during the procedure. This point reinforces the value of explanations on pain level and confirms that pain induced by DTs was not the major determinant of the pain level at end of the procedure. Spending more time with patients to explain the procedure and to control the understanding of our explanation is important to decrease their pain. Some others parameters could be analysed to reduce pain like shock waveform.16 There few data concerning pain and defibrillator in the literature. The majority of DTs are still done under general anaesthesia even in recent publications.13,17 Most of the publications are about pain during shocks in follow-up after implantation or pain in the post-operative period.19,20
There are no recommendations or guidelines concerning anaesthesia during implantation or DT. Some publications referred to conditions of implantation. One publication compared general anaesthesia and LA with sedation and injection of propofol for induction and shock (equivalent protocol to our SGA group).21 Another publication evaluated the defibrillation under LA but with midazolam injection for induction and DT: there were 37 patients in this study and no control group: 70% of patients were symptoms free and most of the remaining patients reported minor symptoms. The pain was evaluated only in four categories: no pain, mild, moderate, and severe pain.22
Concerning the induction, the mode of sedation has no influence in VF induction.
Implanting a defibrillator under MS, according to our results, could have one important consequence:
There were more ineffective first 20 J defibrillation shocks in the MS group (15%) than in the SGA group (2%).
In this group of patients with high defibrillation threshold some parameters were investigated to explain this situation (age of patients, underlying heart disease, ejection fraction, NYHA class, single or bicoil lead, biventricular ICD, type of sedation for induction test: SGA or MS, and longer of VF). Only the duration of VF, the MS group, and biventricular ICD were predictors of higher defibrillation threshold.
Inability to defibrillate 2–15% of the patients with 20 J shock via an ICD seems similar than previously reported.23,24 The influence of anaesthetic choice on defibrillation threshold was investigated by different studies. In animal models, it has been shown that no significant differences in defibrillation threshold were found between animals anaesthetized with thiopental, isoflurane, or halothane.25 In a human study, the authors showed that the anaesthetic choice has no influence on defibrillation threshold in patients undergoing ICD placement under propofol or isoflurane anaesthesia.26 Not only the choice but also the respective dosage of the drugs used for maintenance of anaesthesia was not found to be correlated with the energy used for terminating induced VF.27 Longer VF duration in the MS group than in the SGA group (16.6 ± 4.7 vs. 8.8 ± 3.7 seconds) could increase the defibrillation threshold in the MS group.17 This difference is due to the fact that we waited for that the patient became unconscious before delivering the shock. Therefore, in case of two ineffective tests under MS, we recommend to re-test under general anaesthesia before changing lead's location. This observation can be taken into account in the selection of the type of sedation when the clinician thinks that a patient might have a high defibrillation threshold, for example, in myocardial hypertrophy.28
However, in 80% of cases, the patient was induced only one time, with no problem, no pain, and a good defibrillation threshold. The procedure was simpler with no general anaesthesia.
Study limitations
This study was not randomized but the inclusion group depended only on the day of hospital admission. Evaluation of pain is always difficult. In this study, we used a simple tool, usable for all patients.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Implantable cardioverter-defibrillator implantation under MS even for DT is feasible in the great majority of patients. To reduce pain we need to prefer alternating current to direct current and we need to spend more time to explain the different steps of the implantation procedure.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
[1] Moss AJ, Zareba W, Hall WJ, Klein H, Wilber DJ, Cannom DS, et al. Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction. N Engl J Med (2002) 346:877–83.
[2] Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R, et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure. N Engl J Med (2005) 352:225–37.
[3] Lipscomb KJ, Linker NJ, Fitzpatrick AP. Subpectoral implantation of a cardioverter defibrillator under local anaesthesia. Heart (1998) 79:253–5.
[4] Michaloudis D, Fraidakis O, Kanoupakis E, Flossos A, Manios El. Idiopathic prolonged QT interval and QT dispersion: the effects of propofol during implantation of cardioverter-defibrillator. Eur J Anaesthesiol (1999) 16:842–7.[CrossRef][Web of Science][Medline]
[5] Stix G, Anvari A, Podesser B, Pernerstorfer T, Mayer C, Laufer G, et al. Local anaesthesia versus general anaesthesia for cardioverter-defibrillator implantation. Wien Klin Wochenschr (1999) 111:406–9.[Web of Science][Medline]
[6] Manolis AS, Maounis T, Vassilikos V, Chiladakis J, Cokkinos DV. Electrophysiologist-implanted transvenous cardioverter defibrillators using local versus general anesthesia. Pacing Clin Electrophysiol (2000) 23:96–105.[CrossRef][Medline]
[7] Manolis AS, Vassilikos V, Maounis T, Chiladakis J, Cokkinos DV. Transvenous defibrillator systems implanted by electrophysiologists in the catheterization laboratory. Clin Cardiol (1997) 20:117–24.[Web of Science][Medline]
[8] Raber M, Scattoni V, Roscigno M, Rigatti P, Montorsi F. Perianal and intrarectal anaesthesia for transrectal biopsy of the prostate: a prospective randomized study comparing lidocaine-prilocaine cream and placebo. BJU Int (2005) 96:1264–7.[CrossRef][Web of Science][Medline]
[9] Galosi AB, Minardi D, Dell'atti L, Yehia M, Muzzonigro G. Tolerability of prostate transrectal biopsies using gel and local anesthetics: results of a randomized clinical trial. J Endourol (2005) 19:738–43.[CrossRef][Web of Science][Medline]
[10] Chen YT, Hsiao PJ, Wong WY, Wang CC, Yang SS, Hsieh CH. Randomized double-blind comparison of lidocaine gel and plain lubricating gel in relieving pain during flexible cystoscopy. J Endourol (2005) 19:163–6.[CrossRef][Web of Science][Medline]
[11] Choinière M, Amsel R. A visual analogue thermometer for measuring pain intensity. Journal of pain and symptom management (1996) 11:299–311.[CrossRef][Web of Science][Medline]
[12] Benhamou D. Evaluation de la douleur postopératoire. Ann Fr Anesth Reanim (1998) 17:555–72.[Web of Science][Medline]
[13] Russo AM, Sauer W, Gerstenfeld EP, Hsia HH, Lin D, Cooper JM, et al. Defibrillation threshold testing: is it really necessary at the time of implantable cardioverter-defibrillation insertion? Heart Rhythm (2005) 2:456–61.[CrossRef][Web of Science][Medline]
[14] Boodhoo L, Mitchell A, Ujhelyi M, Sulke N. Improving the acceptability of the atrial defibrillator: patient-activated cardioversion versus automatic night cardioversion with and without sedation (ADSAS 2). Pacing Clin Electrophysiol (2004) 27:910–7.[CrossRef][Medline]
[15] Sulke N, Sayers F, Lip GY. Rhythm control and cardioversion. Heart (2007) 93:29–34.
[16] Boriani G, Biffi M, Silvestri P, Martignani C, Valzania C, Diemberger I, et al. Mechanisms of pain associated with internal defibrillation shocks: results of a randomised study of shock wavesform. Heart Rhythm (2005) 2:708–13.[CrossRef][Web of Science][Medline]
[17] Gradaus R, Bode-Schnurbus L, Weber M, Rotker J, Hammel D, Breithardt G, et al. Effect of ventricular fibrillation duration on the defibrillation threshold in humans. Pacing Clin Electrophysiol (2002) 25:14–9.[CrossRef][Medline]
[18] Jinnouchi Y, Kawahito S, Kitahata H, Tanaka K, Nozaki J, Oshita S. Anesthetic management of a patient undergoing cardioverter defibrillator implantation: usefulness of transesophageal echocardiography and near infrared spectroscopy. J Anesth (2004) 18:220–3.[Medline]
[19] Bollmann A, Marx A, Sathavorn C, Mera F, DeLurgio D, Walter PF, et al. Patient discomfort following pectoral defibrillator implantation using conscious sedation. Pacing Clin Electrophysiol (1999) 22:212–5.[CrossRef][Medline]
[20] Baumert J, Schmitt C, Ladwig KH. Psychophysiologic and affective parameters associated with pain intensity of cardiac cardioverter defibrillator shock discharges. Psychosom Med (2006) 68:591–7.
[21] Schmitt C, Alt E, Plewan A, Schomig A. Initial experience with implantation of internal cardioverter/defibrillators under local anaesthesia by electrophysiologists. Eur Heart J (1996) 17:1710–6.
[22] Stix G, Anvari A, Grabenwoger M, Merhaut C, Turel Z, Laufer G, et al. Implantation of a unipolar cardioverter/defibrillator system under local anaesthesia. Eur Heart J (1996) 17:764–8.
[23] Higgins S, Mann D, Calkins H, Estes NA, Strickberger SA, Breiter D, et al. One conversion of ventricular fibrillation is adequate for implantable cardioverter-defibrillator implant: an analysis from the Low Energy Safety Study (LESS). Heart Rhythm (2005) 2:117–22.[CrossRef][Web of Science][Medline]
[24] Kizilbash AM, Grayburn PA, Anand P, Kowal RC, Page RL, Smith K, et al. Relation of echocardiographic wall motion score index and response to dobutamine stress to defibrillation threshold at the time of implantation of a cardiac defibrillator. Am J Cardiol (2005) 95:1099–101.[CrossRef][Web of Science][Medline]
[25] Jarvis AS, Lahtinen SP. A pilot study: defibrillation thresholds in dogs are similar with isoflurane, halothane, and pentobarbital. Pacing Clin Electrophysiol (1994) 17:280–5.[CrossRef][Medline]
[26] Moerman A, Herregods L, Tavernier R, Jordaens L, Struys M, Rolly G. Influence of anaesthesia on defibrillation threshold. Anaesthesia (1998) 53:1156–9.[CrossRef][Web of Science][Medline]
[27] Camci E, Koltka K, Sungur Z, Karadeniz M, Yavru A, Pembeci K, et al. Implantable cardioverter-defibrillator placement in patients with mild-to- moderate left ventricular dysfunction: hemodynamics and recovery profile with two different anesthetics used during deep sedation. J Cardiothorac Vasc Anesth (2003) 17:613–6.[CrossRef][Web of Science][Medline]
[28] Jastrzebski M, Czarnecka D, Bacior B, Petkow-Dimitrow P, Wilczek R, Kawecka-Jaszcz K. Massive myocardial hypertrophy in hypertrophic cardiomyopathy: a risk factor for sudden cardiac death and high defibrillatin threshold during cardioverter-defibrillator implantation. Kardiol Pol (2005) 63:191–5.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||