Labetalol Versus Magnesium Sulfate (MgSO4) for the Prevention of Eclampsia Trial (LAMPET)

Eclampsia is a major cause of perinatal morbidity and mortality. The pathophysiology is not known but magnetic resonance imaging (MRI) and Doppler data suggest that overperfusion of the cerebral tissues is a major etiologic factor. Hypertensive encephalopathy from overperfusion, and vascular damage from excessive arterial pressure (cerebral barotrauma) are believed to lead to vasogenic and cytotoxic cerebral edema, with resultant neuronal anomalies, seizure activity and cerebral bleeding if left unchecked. Doppler data have shown that cerebral perfusion pressure (CPP) is abnormally increased in severe preeclampsia and that autoregulation of the middle cerebral artery is affected by this condition leading to increased CPP. Magnesium sulfate (MgSO4) is the most widely accepted eclampsia treatment and prophylactic agent, and it has been used in the USA since the 1950's. Despite widespread use, its mechanism of action is unknown. MgSO4 is given intravenously or intramuscularly and requires specialized nursing training and monitoring to minimize toxicity from respiratory and cardiac depression. Labetalol, a combined alpha and beta blocker, has been used for many years to safely treat hypertension in preeclamptic women, and is now known to reduce CPP in women with preeclampsia. In the United Kingdom labetalol was for many years used as the sole agent in treating preeclampsia, and the rate of seizure was no different to that reported in the USA with MgSO4. Since labetalol can be administered orally, is economical, has low toxicity potential, does not require specialized training to administer or monitor, and decreases CPP, it may be an ideal agent for controlling blood pressure (BP) and decreasing the incidence of eclampsia in women with preeclampsia. The current study is a multicenter, randomized, controlled trial to compare the anti-seizure effect of parenteral MgSO4 versus oral labetalol in hypertensive pregnant women who are eligible for MgSO4 therapy. The primary outcome measure is eclampsia, and the secondary outcome measures include blood pressure control, and relevant antenatal, intrapartum, and postnatal maternal and fetal/neonatal parameters including adverse effects and complications. Inclusion criteria are deliberately broad in order to make the study clinically relevant. Hypertensive pregnant women, in whom the decision for delivery has been made, will be enrolled after written, informed consent. Patients will be randomized to receive MgSO4 therapy as given in their institution, versus oral labetalol (200mg/q6 hours), from enrollment in the study until 24 hours post delivery. There will be 4000 patients in each arm of the study and analysis will be by intention-to-treat. The study is powered to show both therapeutic superiority as well as clinical equivalence. This study has the potential to change the way preeclampsia is managed, and will represent a major advance in terms of the availability and safety of prophylactic therapy, especially in developing nations where MgSO4 is underutilized due to cost constraints.

Research Design and Methods: The study will be a multicenter, randomized, controlled, clinical trial comparing the anti-seizure effect of parenteral and/or oral labetalol (n = 4000) versus parenteral (intravenous or intramuscular) magnesium sulfate (n = 4000) with mild or severe preeclampsia who are deemed to be at sufficient risk to warrant seizure prophylaxis with magnesium sulfate. This allows for a 2% lost to follow-up rate. Patients receiving labetalol will either be given a 20 mg intravenous loading dose IV followed by a 200 mg oral dose every 6 hours from the time of inclusion in the study until 24 hours postpartum, or will receive only the 200 mg oral dose every 6 hours from inclusion. Whether or not an IV dose is necessary will be decided by the clinician in charge. Only patients thought to be at sufficient risk to warrant immediate anti-hypertensive therapy will receive the initial IV dose. Patients randomized to the magnesium sulfate group will receive intravenous magnesium sulfate as a 4 or 6 gram loading dose over 20 minutes followed by a 1 or 2 g/h continuous infusion until 24 hours postpartum. The exact MgSO4 protocol used will be determined by the protocol in use at each institution. We have previously shown in the Nimodipine study (as have others with the MAGPIE study) that there is no difference in seizure rate between the 1g/hr and 2g/hr protocols. Patients on the labetalol arm of the study who convulse will be treated with magnesium sulfate. Patients on the magnesium sulfate arm will continue to receive magnesium sulfate. All patients who convulse will be managed per the unit's standard protocol for eclampsia treatment. Principal Outcome Measure: Occurrence of eclamptic seizure(s) after enrollment in the study. Secondary Outcome Measures Maternal: Blood pressure, pulse pressure and heart rate changes Need for additional antihypertensive medication (defined by need to control BP > 160 mmHg systolic and/or 110 mmHg) Subjective assessment of side effects by the patient Objective assessment of new onset complications and/or side effects by the treating clinicians Labor and delivery parameters including; induction to delivery interval, rate of cervical dilatation, oxytocin dosages, length of labor, delivery route, blood loss at delivery, and postpartum course Type of anesthesia administered (none, local infiltration for delivery only, epidural for labor, epidural for delivery, spinal for delivery, combined/spinal epidural for labor and delivery, general anesthesia for delivery). Fetal and Neonatal: Occurrence of newly diagnosed fetal distress during labor necessitating emergent delivery Umbilical cord gases at delivery (if available in institution); Apgar scores Neonatal outcome as defined by NICU admission, hospital survival to discharge, length of hospital survival, days in NICU, need for blood transfusion, need for pressor agents, need for mechanical ventilation, neonatal cardiac dysrhythmias, neonatal cardiac failure, necrotizing enterocolitis, sudden infant death, neonatal sepsis, neonatal hypoglycemia, and there will be an other block for recording any other identified complications not mentioned. Institutional Review and Informed Consent Procedures: The protocol for this study will be approved by the Institutional Review Board of each participating institution and copies of the approval will be on file with the principal investigator prior to dispensing of the drug. All patients will give written informed consent, and will have the protocol described to them in their native language if they do not understand English. Study Procedures Informed Consent: The final protocol will first be approved by the appropriate IRBs of the US and foreign institutions as well as the data coordinating center. All women must participate in an informed consent dialogue and must give their written informed consent prior to enrollment and participation in the study. Eligible patients that fit the required criteria for being entered into the study, but do not wish to enter into the study or their primary physician doesn't wish them to participate will be recorded. This will allow us to determine the percentage of patients who are eligible but are not enrolled. Enrollment: Once delivery is planned, and inclusion criteria have been satisfied, the patient will either be approached and counseled by an investigator or co-investigator (who may be the patient's primary care physician) and informed written consent will be taken, or she will be informed of the study and given a consent document by the nurse to read and then telephonically consented by an investigator at that site. The nurse will then document on the consent form that the patient was consented by the physician. If the patient is telephonically consented, a site investigator will co-sign the consent document at a later time. Entry into this study will not in any way influence the obstetric management of the patient, and the patient will be informed of her freedom to withdraw at any time without compromise to the standard of her treatment. All routine patient care and routine observations will be carried out by the labor and delivery staff. Each institution in the USA will be cross-referenced to the IND number of the US Principal Investigator (#63,966). This IND number is only necessary for grant funding. The FDA has assessed this study and does not feel that it required an IND number for clinical reasons since labetalol is already an approved antihypertensive medication. The only reason that an IND number was requested was to allow us to satisfy the requirements for various Federal Funding agencies. The foreign institutions will not require an IND number but will all require an active approval from their Institutional Review Board. Study Interventions Interventions: Patients will be assigned to one of two treatment groups with a randomization sequence prepared and maintained centrally by the database. The random permuted block method of randomization will be used because it guarantees that at no time during randomization will the imbalance be large and at certain points the number of patients in each group will be equal. Because the study is unblinded, random block sizes will be used. Randomization will be stratified by clinical center, parity, and age to assure balance between the two treatment groups with respect to anticipated differences in the patient population and possible differences in patient management. Sufficient stocks of the study drugs are kept in the labor and delivery areas of the study sites to ensure no undue delay in beginning the therapy. This will reduce the chance of a seizure occurring after randomization but before drug administration. Since this study will have an intent-to-treat analysis, such patients will still be included in the analysis in the group to which they were assigned. The study drugs will be kept in appropriately secure areas in the labor and delivery suite of each site. This system will allow for the most efficient administration of drug after randomization and will limit the rate of drop out because of seizure prior to getting the intended drug. In this intent-to-treat analysis, each patient will be analyzed in the group she was originally assigned to regardless of whether or not she received the drug or not. All patient data will be recorded in an internet-based interactive computer database. The URL is www.labetalol.org. The database has been designed by MedSciNet (http://www.medscinet.se/msn/home.htm) who have extensive experience in designing, maintaining and managing large multicenter internet based studies for national and international organizations (WHO, Government of Sweden). The database satisfies all HCA, HIPPA, FDA CRF 21 Part 11, and NIH security protocols. Each study site will have a computer available to the research team for this study. The database is set up to ensure that correct and complete data are entered. The program determines eligibility and will only allow enrollment once the specific inclusion and exclusion criteria are satisfied. If eligible a consent form will be offered to the patient. Once she has signed the consent form she will be formally enrolled following the prompts on the screen. The computer will then randomize her to a treatment group. There will be prompts that will guide the research team through the data collection sheets. There are reminders built in to ensure that essential demographic and primary outcome data are entered. In addition certain fields will not allow data entry if other essential fields are not completed or have incorrect data entered. The program will query any impossible numbers that may be entered in error. This will hopefully reduce errors from data transcription and will ensure that the protocol is adhered to. Adverse event datasheets will be automatically displayed and will require completion prior to continuation of the study once any reportable events are noted. These and all other data will always be available to the (Data Safety and Monitoring Board) DSMB for analysis as outlined elsewhere in this proposal. Each PI will be responsible for ensuring the integrity of the data from their site. Prior to the data being stored in the main database a sign-off will be required by each site PI. In addition, the Study Coordinator and Study PI will be responsible for checking all data for quality issues and reliability on a regular basis. The main database will be monitored for accuracy and quality of data by the Principal Investigator at the St. Mark's Hospital site in Salt Lake City, Utah. All investigators will be kept current on any/all adverse event reports and interim analyses. Patients randomized to labetalol will receive two 100 mg tablets orally. The labetalol tablets (200 mg) will be repeated every 6 hours, unless the patient is hypotensive (BP < 90 mmHg systolic and/or 60 mmHg diastolic). The tablets will be swallowed with a maximum of 10cc of water. In those patients where the clinical team desires a more rapid intravenous dosing, a single 20mg of labetalol will be given before the oral dose is taken. Labetalol therapy will be continued for 24 hours post delivery. If the blood pressure is not controlled (reduction in BP to < 160 systolic and < 110 mmHg diastolic) within 40 minutes of beginning labetalol therapy, the patient will receive an intravenous dose of 20mg labetalol, which can be increased in an escalating dose to a maximum of 80mg (20 mg, 40 mg, 80 mg) every 20 minutes as necessary to control the blood pressure until a cumulative maximum of 240 mg has been given. It is extremely important that the blood pressure is controlled as outlined in this protocol and any deviation from these instructions will be deemed a protocol violation. Even if repeated labetalol is required for blood pressure control, the patient will continue to receive 200mg labetalol orally every 6 hours unless she has bradycardia or hypotension (as defined by a blood pressure of < 90 mmHg systolic and/or < 60 mmHg diastolic). If blood pressure is not controlled with labetalol the patient will have her blood pressure controlled by the delivery team using whatever other agent they choose. In those institutions where intravenous labetalol is unavailable the patient will receive an intravenous dose of 5mg hydralazine, which can be repeated every 20 minutes as necessary to control the blood pressure. Patients randomized to the magnesium sulfate arm of the study will receive a 4-6 gram intravenous loading dose per institution protocol over at least 20 minutes, and this will be followed by a continuous infusion of 1-2 grams per hour. Magnesium sulfate levels will be recorded as per the usual protocol of the participating hospital only in those hospitals where this is the protocol. Magnesium sulfate levels are not mandatory as long as there is an active protocol for determining the presence or absence of magnesium sulfate toxicity. Magnesium sulfate therapy will be discontinued 24 hours after delivery. If the blood pressure is not controlled (reduction in BP to below 160 mmHg systolic and 110 mmHg diastolic) within 40 minutes of beginning magnesium sulfate therapy, the patient will be given an intravenous dose of 5mg hydralazine, which can be repeated every 20 minutes as necessary to control the blood pressure. If the patient still needs further BP control she should be given anything EXCEPT labetalol for BP control, i.e. nifedipine, hydralazine, etc. Persistent blood pressure readings above 160 systolic and/or 110 diastolic are a protocol violation. Following randomization, patients will be managed as per the standard labor and delivery protocol for preeclampsia in that institution. The only restrictions will be that no other anti-seizure medications (apart from the study agents) will be given unless the patient is removed from the study. Use of non seizure-related drugs used will be recorded on the case report. The enrollment of patients in the study will not, other than in the choice of the anti-seizure agent, influence the obstetrical management of the patient. Routine hourly vital signs and fetal heart rate monitoring will be employed. A standardized case report form will be used for all study patients. Patients who experience a seizure prior to receiving their study drug, but after being randomized to the study, will be removed from active participation in the study and treated according to routine practice in that institution. Those patients randomized to labetalol and who have a seizure on the drug will be started on magnesium sulfate per institutional protocol. Patients on the magnesium sulfate arm of the study who convulse will be continued on magnesium sulfate per institutional protocol. Since all of the above mentioned patients will have been randomized they will be included in the intent to treat analysis. Outcome data will be obtained on all randomized patients, regardless of whether they receive study drug or have to stop study drug. Data Collection and Storage: All patient data will be recorded in an internet-based interactive computer database. The URL is www.labetalol.org and for the purposes of this grant application a testing/training program can be accessed by reviewers using the URL http://www.medscinet.com/labetaloltest/ the login name: training and the password: training will allow reviewers to review the training web site. The database has been designed by MedSciNet (http://www.medscinet.se/msn/home.htm) who have extensive experience in designing, maintaining and managing large multicenter internet based studies for national and international organizations (WHO, Government of Sweden). The database satisfies all HCA, HIPPA, FDA CRF 21 Part 11, and NIH security protocols. Each study site will have a computer available to the research team for this study. The database is set up to ensure that correct and complete data are entered. The program determines eligibility and will only allow enrollment once the specific inclusion and exclusion criteria are satisfied. If eligible a consent form will be offered to the patient. Once she has signed the consent form she will be formally enrolled following the prompts on the screen. The computer will then randomize her to a treatment group. There will be prompts that will guide the research team through the data collection sheets. There are reminders built in to ensure that essential demographic and primary outcome data are entered. In addition certain fields will not allow data entry if other essential fields are not completed or have incorrect data entered. The program will query any impossible numbers that may be entered in error. This will hopefully reduce errors from data transcription and will ensure that the protocol is adhered to. Adverse event datasheets will be automatically displayed and will require completion prior to continuation of the study once any reportable events are noted. These and all other data will always be available to the DSMB for analysis as outlined elsewhere in this proposal. Each PI will be responsible for ensuring the integrity of the data from their site. Prior to the data being stored in the main database a sign-off will be required by each site PI. In addition, the Study Coordinator and Study PI will be responsible for checking all data for quality issues and reliability on a regular basis. The main database will be monitored for accuracy and quality of data by the Principal Investigator at the St. Mark's Hospital site in Salt Lake City, Utah. All investigators will be kept current on any/all adverse event reports and interim analyses. The Principal Investigator of the study, Dr. Michael A. Belfort, MD, PhD or a designated study co-investigator will be in contact with each of the institutions on a regular basis during the 5 years of the study to monitor the trial. The Trial Coordinator will ensure that all personnel are trained on the system. She will monitor enrollment on a daily basis and will be responsible for ensuring that all data are transmitted to the central database in a timely manner and that data integrity is maintained. Study Timetable Patient Recruitment: One of the major stumbling blocks in any research into the pathophysiology of eclampsia is the enrollment of sufficient patients to achieve statistical significance. No single center is capable of completing a study of the scope and nature of that proposed in a reasonable time period. For this reason it is essential that multicenter networks such as the one that we have established be used to their full capacity and supported adequately to allow this to happen. All of the participating institutions in our network deliver in excess of 1500 patients per year. We have shown at the St. Marks and University of Utah sites that enrollment of 2 patients per week is easily attainable with on-site presence and encouragement. We have asked each site to enroll a minimum of 2 patients per week and we expect more from the larger sites (Texas Women's Hospital - 8000 deliveries per annum, University of Minnesota Medical Center Hospitals 10 000 patients per annum, Utah HCA facilities 8000 deliveries per annum). Given an enrollment of 2 patients per week from 20 sites we anticipate annual enrollment of at least 2000 patients and that this study can be realistically performed within 5 years. In addition, we expect that as the LAMPET Study gains impetus and information is disseminated within the HCA network of obstetric institutions we will have more sites joining the study. There is an active program within the HCA Perinatal Safety Initiative that rewards facilities for Quality Programs. HCA has started a grading system within its obstetric hospitals that has as its acme "Center of Excellence" status. This status within the HCA organization confers significant advantages to the facility and is as such a highly sought after designation. One of the components for attaining Center of Excellence status is an active obstetric research program, and we expect that the LAMPET study will be established and ongoing, it will represent an attractive option for centers wishing to gain the Center of Excellence goal. As outlined above there are already a number of the larger HCA facilities that have committed a 0.5 nursing FTE to this project and have been included in the grant application. There are more facilities that have pledged support and committed nursing support but are not eligible for grant funding because of their smaller size and lower potential for enrolling the required 2 patients per week. Despite the fact that they will not be paid anything, their CEO's have committed up to a 0.5 nursing FTE to help these smaller facilities participate. We anticipate that more and more of the smaller hospitals will also join the study as we disseminate the project throughout the company at divisional meetings and national conferences. These particular sites are prepared to participate in this study without any support and we anticipate that as the study progresses and the protocol is disseminated further HCA facilities will sign up simply for the Center of Excellence designation. This will further ensure success of this project. HCA has a Letter of Agreement program that allows any hospital within the system to accept the IRB approval from the St. Marks site, and thus circumvents time-consuming and costly bureaucracy. This will allow individual physicians and groups of practitioners who wish to participate in research to play an active role in a major RCT. Potential Principal Investigators will be able to download the protocol and a template consent from the Study Website for their own use. All consent form and protocol changes that may be made will be vetted by Dr. Belfort to ensure that these are only cosmetic to satisfy the particular needs of that institution. No substantive changes to the protocol or consent form will be allowed. Timetable: 5 years will be a realistic amount of time to successfully complete the study. Statistical Analysis: One of the most widely used procedures for interim analysis of a clinical trial is the spending function approach of Lan and DeMets. This method uses a function of time to specify the rate at which the total Type I error probability is to be spent during the trial so that this quantity does not exceed the desired alpha level. For this study, the spending function corresponding to an O'Brien and Fleming boundary will be used. With this method, the rate at which the type I error is spent is a function of the fraction of total information available at the time of the interim analysis (i.e., information time). The information time is determined by the type of statistical test employed. For the analysis of seizure rate, the information time is the proportion of the total patients randomized. The table below provides the Z score at each time point required for considering terminating the study early and for final analysis to show statistical significance at the alpha=0.05 level: The Lan-DeMets procedure is flexible in that the total number and timing of the sequential analyses need not be specified in advance. Also, there is no requirement that looks be equally spaced. In some clinical trials, however, a surprising development may lead to extra "looks" or even continuous monitoring of data. The Lan-DeMets procedure is flexible, in that it can switch from occasional to continuous monitoring of the data with negligible effect on the Type I error level. Data Monitoring Committee: An independent data monitoring committee has been established to review the progress of the study. The data monitoring committee will meet periodically to review trial results, monitor patient safety and study progress. The data monitoring committee will meet when approximately 25%, 50% and 75% of the patients have been enrolled and delivered. However, this committee will be responsible for deciding whether more or less frequent meetings are necessary. The data monitoring committee will be completely independent of the study group and will not include any member of the research team or any individual with a conflict of interest.

Preeclampsia, Pregnancy Induced Hypertension, Gestational Hypertension, Chronic Hypertension, Superimposed Preeclampsia

20mg IV (can be increased to an escalating dose of maximum of 80 mg (20mg, 40 mg, 80 mg q 20min) Cumulative max of 240 mg. 20mg labetalol PO every 6 hours

Need for additional antihypertensive medication (defined by need to control BP > 160 mmHg systolic and/or 110 mmHg)

Labor and delivery parameters including; induction to delivery interval, rate of cervical dilatation, oxytocin dosages, length of labor, delivery route, blood loss at delivery, and postpartum course; and

Type of anesthesia administered (none, local infiltration for delivery only, epidural for labor, epidural for delivery, spinal for delivery, combined/spinal epidural for labor and delivery, general anesthesia for delivery).

Fetal and Neonatal: Occurrence of newly diagnosed fetal distress during labor necessitating emergent delivery

Neonatal outcome as defined by NICU admission, hospital survival to discharge, length of hospital survival, days in NICU, need for blood transfusion, need for pressor agents, need for mechanical ventilation, neonatal cardiac dysrhythmias,

neonatal cardiac failure, necrotizing enterocolitis, sudden infant death, neonatal sepsis, neonatal hypoglycemia, and there will be an other block for recording any other identified complications not mentioned.

Inclusion Criteria: Any patient with preeclampsia (BP > 140 systolic and/or > 90 mmHg diastolic with 1+ or more proteinuria [or a 24 hour specimen with > 300 mg/day]), chronic hypertension (or superimposed preeclampsia), or gestational hypertension deemed to be at risk for eclamptic convulsions and who would routinely be treated in the participating institution with some form of anti-seizure prophylaxis during labor and delivery. Exclusion Criteria: Any patient for whom informed consent cannot be obtained. Any patient who has received an antihypertensive medication within 6 hours prior to enrollment will not be eligible but those who have received antihypertensive medications other than beta-blockers or magnesium sulfate may still be enrolled as long as they have not been given a dose within the 6 hours prior to enrollment. If a patient has received MgSO4 or a short acting beta-blocker or calcium channel blocker more than 12 hours prior to enrollment or if they have received a long acting beta-blocker more than 24 hours before enrollment she may still be considered eligible. This stipulation will allow increased recruitment of patients especially those with chronic hypertension and those transferred from outlying institutions. We expect these patients to be a minority of the enrollment. A history of bronchial asthma, emphysema, heart block, angina, cardiomyopathy or myocardial infarction. Any history or signs of congestive cardiac failure, or arrhythmia with a ventricular rate of less than 60 bpm. Patients with severe mental or physical disorders which, in the opinion of the investigators, might affect responsiveness to therapy or any other aspect of the study. Patients who are allergic to drugs with a chemical structure similar to labetalol or magnesium sulfate. Patients given magnesium sulfate, labetalol or short acting beta blockers or calcium channel blockers less than 12 hours prior to enrollment in the study. Evidence of fetal distress or fetal anomalies. Inability to secure intravenous access. Patient's primary physician declines to enroll patient in study.

Belfort MA, Anthony J, Saade GR, Allen JC Jr; Nimodipine Study Group. A comparison of magnesium sulfate and nimodipine for the prevention of eclampsia. N Engl J Med. 2003 Jan 23;348(4):304-11. doi: 10.1056/NEJMoa021180.

Belfort MA. Is high cerebral perfusion pressure and cerebral flow predictive of impending seizures in preeclampsia? A case report. Hypertens Pregnancy. 2005;24(1):59-63. doi: 10.1081/PRG-45776.

Belfort MA, Tooke-Miller C, Allen JC Jr, Dizon-Townson D, Varner MA. Labetalol decreases cerebral perfusion pressure without negatively affecting cerebral blood flow in hypertensive gravidas. Hypertens Pregnancy. 2002;21(3):185-97. doi: 10.1081/PRG-120015845.

Belfort MA, Tooke-Miller C, Varner M, Saade G, Grunewald C, Nisell H, Herd JA. Evaluation of a noninvasive transcranial Doppler and blood pressure-based method for the assessment of cerebral perfusion pressure in pregnant women. Hypertens Pregnancy. 2000;19(3):331-40. doi: 10.1081/prg-100101995. Erratum In: Hypertens Pregnancy 2001;20(1):139-40.

Riskin-Mashiah S, Belfort MA. Cerebrovascular hemodynamics in chronic hypertensive pregnant women who later develop superimposed preeclampsia. J Soc Gynecol Investig. 2005 Jan;12(1):28-32. doi: 10.1016/j.jsgi.2004.08.002.

Belfort MA, Varner MW, Dizon-Townson DS, Grunewald C, Nisell H. Cerebral perfusion pressure, and not cerebral blood flow, may be the critical determinant of intracranial injury in preeclampsia: a new hypothesis. Am J Obstet Gynecol. 2002 Sep;187(3):626-34. doi: 10.1067/mob.2002.125241.

Belfort MA, Saade GR, Yared M, Grunewald C, Herd JA, Varner MA, Nisell H. Change in estimated cerebral perfusion pressure after treatment with nimodipine or magnesium sulfate in patients with preeclampsia. Am J Obstet Gynecol. 1999 Aug;181(2):402-7. doi: 10.1016/s0002-9378(99)70569-7.