The effect of pre-eclampsia on retinal microvascular caliber at delivery and post-partum.

BACKGROUND: The retinal microcirculation provides a unique view of microvessel structure by means of non-invasive, retinal image analysis. The aim of the study was to compare the retinal vessel caliber at delivery and one-year post-partum between women who have had pre-eclampsia during pregnancy to a normotensive control group. METHODS: Digital photos of the eye were taken at delivery and one-year post-partum. Retinal vessels were analysed and summarised as the corrected central retinal arteriolar equivalent and corrected central retinal venular equivalent. RESULTS: The corrected central retinal arteriolar equivalent and corrected central retinal venular equivalent were significantly lower in the pre-eclamptic group compared to the control group both at delivery and one-year post-partum (p < 0.001). CONCLUSION: Retinal artery and venular caliber changes that occur during pregnancies affected by pre-eclampsia persist for up to one-year post-partum. These changes may reflect a permanent, long-term microvascular dysfunction and may be useful as a biomarker of future vascular risk.

Cerebral white matter lesions after pre-eclampsia.

BACKGROUND: Women who have had pre-eclampsia in their previous pregnancies demonstrate a greater prevalence of cerebral white matter lesions several years after the pregnancy than women who have been normotensive during their pregnancy. Both the pathophysiology and the timing of development of these lesions are uncertain. White matter lesions, in the general population, are associated with an increased risk of stroke, dementia and death. AIMS AND OBJECTIVES: The objective of the study was to determine the prevalence of cerebral white matter lesions amongst women with severe pre-eclampsia at delivery, 6months and 1year postpartum and to establish the possible pathophysiology and risks factors. METHODS: This was a longitudinal study performed at Steve Biko Academic Hospital, a tertiary referral hospital in Pretoria South Africa. Ninety-four women with severe pre-eclampsia were identified and recruited during the delivery admission. Magnetic resonance imaging (MRI) of the brain was performed post - delivery and at 6months and 1year postpartum. RESULTS: Cerebral white matter lesions were demonstrated in 61.7% of women at delivery, 56.4% at 6months and 47.9% at 1year. Majority of the lesions were found in the frontal lobes of the brain. The presence of lesions at 1year post-delivery was associated with the number of drugs needed to control blood pressure during pregnancy (OR 5.1, 95% CI 2.3-11.3, p<0.001). The prevalence of WMLs at 1year was double in women with chronic hypertension at 1year compared to those women who were normotensive (65.1% vs 32.3%). CONCLUSION: Women who require 2 or more drugs to control blood pressure during pregnancy have an increased risk of developing cerebral white matter lesions after delivery.

A randomised trial comparing laparoscopy with laparotomy in the management of women with ruptured ectopic pregnancy.

BACKGROUND: Ruptured ectopic pregnancy (REP) is a common gynaecological emergency in resource-poor settings, where laparotomy is the standard treatment despite laparoscopic surgery being regarded as the optimal treatment. There is a lack of prospective randomised data comparing laparoscopic surgery with laparotomy in the surgical management of women with REP. OBJECTIVE: To compare operative laparoscopy with laparotomy in women with REP. METHODS: This was a randomised parallel study. One hundred and forty women with suspected REP were randomised to undergo operative laparoscopy or laparotomy. The outcome measures were operating time, hospital stay, pain scores and analgesic requirements, blood transfusion, time to return to work, and time to full recovery. RESULTS: Operating time was significantly longer in the laparoscopy group (67.3 v. 30.5 minutes, p<0.001). Duration of hospital stay, pain scores and need for analgesia were significantly less in the laparoscopy group. Women in this group returned to work 8 days earlier and their time to full recovery was significantly shorter compared with those in the laparotomy group. Significantly more women undergoing laparotomy required blood transfusion than women in the laparoscopy group. In the latter group, 14.5% of women required blood transfusion compared with 26.5% in the laparotomy group (p=0.01). CONCLUSION: Operative laparoscopy in women treated for REP is feasible in a resource-poor setting and is associated with significantly less morbidity and a quicker return to economic activity.

A randomised trial comparing laparoscopy with laparotomy in the management of women with ruptured ectopic pregnancy

Background. Ruptured ectopic pregnancy (REP) is a common gynaecological emergency in resource-poor settings, where laparotomy is the standard treatment despite laparoscopic surgery being regarded as the optimal treatment. There is a lack of prospective randomised data comparing laparoscopic surgery with laparotomy in the surgical management of women with REP.Objective. To compare operative laparoscopy with laparotomy in women with REP.Methods. This was a randomised parallel study. One hundred and forty women with suspected REP were randomised to undergo operative laparoscopy or laparotomy. The outcome measures were operating time, hospital stay, pain scores and analgesic requirements, blood transfusion, time to return to work, and time to full recovery.Results. Operating time was significantly longer in the laparoscopy group (67.3 v. 30.5 minutes, p<0.001). Duration of hospital stay, pain scores and need for analgesia were significantly less in the laparoscopy group. Women in this group returned to work 8 days earlier and their time to full recovery was significantly shorter compared with those in the laparotomy group. Significantly more women undergoing laparotomy required blood transfusion than women in the laparoscopy group. In the latter group, 14.5% of women required blood transfusion compared with 26.5% in the laparotomy group (p=0.01). Conclusion. Operative laparoscopy in women treated for REP is feasible in a resource-poor setting and is associated with significantly less morbidity and a quicker return to economic activity

Basic and comprehensive emergency obstetric and neonatal care in 12 South African health districts.

AIM: To assess the functionality of healthcare facilities with respect to providing the signal functions of basic and comprehensive emergency obstetric care in 12 districts. SETTING: Twelve districts were selected from the 52 districts in South Africa, based on the number of maternal deaths, the institutional maternal mortality ratio and the stillbirth rate for the district. METHODS: All community health centres (CHCs) and district, regional and tertiary hospitals were visited and detailed information was obtained on the ability of the facility to perform the basic (BEmONC) and comprehensive (CEmONC) emergency obstetric and neonatal care signal functions. RESULTS: Fifty-three CHCs, 63 district hospitals (DHs), 13 regional hospitals and 4 tertiary hospitals were assessed. None of the CHCs could perform all seven BEmONC signal functions; the majority could not give parenteral antibiotics (68%), perform manual removal of the placenta (58%), do an assisted delivery (98%) or perform manual vacuum aspiration of the uterus in a woman with an uncomplicated incomplete miscarriage (96%). Seventeen per cent of CHCs could not bag-and-mask ventilate a neonate. Less than half (48%) of the DHs could perform all nine CEmONC signal functions (81% could perform eight of the nine functions), 24% could not perform caesarean sections, and 30% could not perform assisted deliveries. CONCLUSIONS: The ability of the CHCs and district hospitals to perform the signal functions (lifesaving services) of basic and comprehensive emergency obstetric care was poor in many of the districts studied. This implies that safe maternity care was not consistently available at many facilities conducting births.

Hyperferritinemia and markers of inflammation and oxidative stress in the cord blood of HIV-exposed, uninfected (HEU) infants.

OBJECTIVES: The purpose of this study was to evaluate markers of iron status and inflammation/oxidative stress in maternal and cord blood (CB) of HIV-infected and HIV-uninfected women as potential mechanisms for poor outcomes among HIV-exposed, uninfected (HEU) infants. METHODS: Maternal venous blood and CB specimens were obtained from 87 pregnant women (45 HIV-infected and 42 HIV-uninfected) enrolled at Kalafong Hospital, Pretoria, South Africa. Iron status [serum iron, ferritin and transferrin concentrations, transferrin saturation, soluble transferrin receptor (sTfR) concentration and the sTfR/log ferritin (sTfR/F) index], antenatal exposure to inflammation (CB C-reactive protein and interleukin-6 concentrations and haptoglobin switch-on status) and oxidative stress [total radical trapping ability of CB plasma (TRAP) and chronic oxidative stress (soluble receptor of advanced glycation end-products (sRAGE) concentration] were assessed in laboratory studies. RESULTS: There were no differences between the HIV-infected and HIV-uninfected groups in maternal haematological and iron indices, except that HIV-infected mothers had decreased white blood cell counts (P = 0.048) and increased serum ferritin concentrations (P = 0.032). Ferritin levels were significantly higher in CB than in maternal blood (P < 0.001) in both groups and further elevated in the CB of HEU infants (P = 0.044). There was also an inverse relationship between CB sTfR/F index and sRAGE (r = -0.43; P = 0.003) in the HIV-infected but not in the HIV-uninfected group. CONCLUSIONS: Our study showed for the first time that ferritin was significantly elevated in CB of HEU infants. The inverse relationship between sTfR/F index and sRAGE in CB suggests that chronic oxidative stress or RAGE axis activation in HIV-infected mothers may play a role in modulating ferritin levels.

Support group processes: Perspectives from HIV-infected women in South Africa.

This study examined the experiences and perceived benefits of support group participation among HIV-infected women in South Africa. From a qualitative analysis of responses, key psychological processes through which support groups are potentially beneficial were identified. These processes included: identification; modeling; acceptance; and empowerment. The participants' consequent life changes were explored in order to associate these processes with the positive outcomes of support group participation. Through understanding the relationship between the psychological processes within a support group setting and the potential benefits, and by targeting these processes in the development and implementation of future support group interventions, a framework is provided for achieving positive outcomes associated with support group participation.

HIV/AIDS stigma in a South African community.

HIV/AIDS-related stigma threatens to undermine interventions to prevent and treat HIV/AIDS. To address stigma in a South African community, a thorough understanding of the nature of stigma in the specific cultural context is needed. The goals of this research were to assess the level of stigmatising attitudes among members of a community, compare this to the level of stigma that is perceived to exist within the community and determine to what extent stigmatising attitudes are affected by socio-demographic characteristics, HIV-related experience and cultural beliefs. A questionnaire was completed by 1077 respondents in key areas in two communities in Tshwane, South Africa. The questionnaire included an assessment of HIV-related experience, HIV-knowledge, personal stigma and perceptions of stigma within the community. The findings indicate that the level of personal stigma was significantly lower than that perceived to be present in the community. Respondents who were more stigmatising were older, male, less educated and less knowledgeable about HIV. They were less likely to know someone with HIV and had more traditional cultural viewpoints. While socio-demographic and cultural factors are difficult to change, efforts aimed at increasing people's knowledge and experience of the epidemic occurring in their community could change the level of stigmatising attitudes within their community. Such efforts could have potential benefits in addressing the epidemic and providing greater support for those with HIV.

Can the outcome of induction of labour with oral misoprostol be predicted.

OBJECTIVE: To determine predictors of outcome for induction of labour using oral misoprostol. SETTING: Labour ward at Kalafong Hospital in Atteridgeville, Pretoria, that serves an indigent South African urban population. METHODS: Data were collected prospectively on all women undergoing induction of labour with oral misoprostol from 1 March 2004 to 28 February 2005. Patients with contraindications to misoprostol induction were excluded. Univariate analysis and logistical regression analysis were performed to determine the significant predictors of success of induction of labour. Successful induction was defined as a vaginal delivery achieved within 24 hours. RESULTS: Five hundred and fifty-eight patients were included. There were three major indications for induction of labour, namely hypertension (45%), postdates (22.1%) and prelabour rupture of membranes (20.6%). Vaginal delivery was achieved within 24 hours in 52.4% of patients. The caesarean section rate was 42.1%. Fetal heart rate changes occurred in 25.6% and hyperstimulation in 1.4% of patients. Logistical regression analysis identified the following parameters as independent predictors of vaginal delivery achieved within 24 hours: primiparity (p < 0.001), Bishop score < 3 (p < 0.001), Bishop score 4 - 6 (p = 0.029), ruptured membranes (p < 0.001) and pre-eclampsia (p = 0.006). A method of scoring (Mbele score) has been developed making use of the results of this analysis in order to predict the successful outcome of induction. CONCLUSIONS: Primigravidity, intact membranes, pre-eclampsia and a low Bishop score were indicators of an unsuccessful outcome for induction of labour. It is thought that the Mbele score will be helpful in counselling patients on methods of delivery when they are admitted for induction of labour.

Critical incident audit and feedback to improve perinatal and maternal mortality and morbidity.

BACKGROUND: Audit and feedback of critical incidents is an established part of obstetric practice. However, the effect on perinatal and maternal mortality is unclear. The potential harmful effects and costs are unknown. OBJECTIVES: Is critical incident audit and feedback effective in reducing the perinatal mortality rate, the maternal mortality ratio, and severe neonatal and maternal morbidity? SEARCH STRATEGY: We searched the Cochrane Pregnancy and Childbirth Group Trials Register (January 2005), the Cochrane Effective Practice and Organisation of Care Group Trials Register (January 2005), MEDLINE (1965 to December 2004), EMBASE (1965 to December 2004), SCIBASE (1965 to December 2004) and the World Health Organization systematic review of maternal mortality and morbidity database (January 1997 to December 2002). SELECTION CRITERIA: Randomized trials of audit (defined as any summary of clinical performance over a specified period of time) and feedback (method of feeding that information back to the clinicians) that reported objectively measured professional practice in a healthcare setting or healthcare outcomes. DATA COLLECTION AND ANALYSIS: No suitable trials were found. MAIN RESULTS: None. AUTHORS' CONCLUSIONS: The necessity of recording the number and cause of deaths is not in question. Mortality rates are essential in identifying problems within the healthcare system. Maternal and perinatal death reviews should continue to be held, until further information is available. The evidence from serial data clearly suggests more benefit than harm. Feedback is essential in any audit system. The most effective mechanisms for this are unknown, but it must be directed at the relevant people.

Aggressive or expectant management of labour: a randomised clinical trial.

OBJECTIVE: To compare labour outcomes using aggressive or expectant management protocols. DESIGN: Randomised trial. SETTING: Pretoria Academic Complex, South Africa. It serves an indigent urban population. POPULATION: Healthy nulliparous women in active labour, at term, with a health singleton pregnancy and cephalic presentation. METHODS: The women were randomised to either aggressive (n = 344) or expectant (n = 350) management protocols. Aggressive management entailed using a single line partogram, a vaginal examination every two hours and use of an oxytocin infusion if the line was crossed. Expectant management entailed using a two line partogram, with the alert line and a parallel action line four hours to the right, with a vaginal examination every four hours. If the action line was reached, oxytocin was started. The women were reassessed every two hours thereafter. Analgesia was prescribed on request. MAIN OUTCOME MEASURES: Mode of birth, use of oxytocin and analgesia and neonatal outcome. RESULTS: The groups were similar with respect to maternal age, cervical dilation at trial entry, number crossing the alert line and birthweight of the infants. Significantly fewer women managed aggressively had caesarean sections (16.0%) than those managed expectantly (23.4%) (relative risk [RR] 0.68, 95% confidence intervals [CI] 0.50, 0.93). Significantly more oxytocin was used in the aggressive management group, but there was no difference with respect to the use of analgesia or episiotomy or in neonatal outcome with respect to the Apgar score at 1 or 10 minutes. There were three perinatal deaths. One woman was found to have an intrauterine death before trial entry and the other two were in the aggressive management group but did not receive oxytocin. Compliance by staff was poor in the aggressive management group. CONCLUSIONS: Aggressive management of labour reduces the caesarean section rate in nulliparous women but requires more intensive nursing.

The use of dexamethasone in women with preterm premature rupture of membranes--a multicentre, double-blind, placebo-controlled, randomised trial. Dexiprom Study Group.

OBJECTIVE: To assess whether administration of dexamethasone in women with preterm premature rupture of membranes (PPROM) has an effect on the prevalence of maternal sepsis, neonatal respiratory distress syndrome (RDS), perinatal mortality and neonatal sepsis in a developing country. SETTING: Six public hospitals in South Africa that deal mainly with indigent women. METHOD: A multicentre, double-blind, placebo-controlled, randomised trial was performed on women with PPROM and fetuses of 28-34 weeks' gestation or clinically estimated fetal weight between 1,000 and 2,000 g if the gestational age was unknown. Women were randomised to receive either dexamethasone 24 mg intramuscularly or placebo in two divided doses 24 hours apart. All women received amoxycillin and metronidazole and were managed expectantly. Hexoprenaline was administered if contractions occurred within the first 24 hours after admission to the trial. OUTCOME MEASURES: The maternal outcome measures were clinical chorio-amnionitis and postpartum sepsis. The outcome measures for infants were perinatal death, RDS, mechanical ventilation, necrotising enterocolitis, and neonatal infection within 72 hours. RESULTS: One hundred and two women who delivered 105 babies were randomised to the dexamethasone group and 102 women who delivered 103 babies, to the placebo group. The groups were well balanced with regard to clinical features. There was a trend towards fewer perinatal deaths in the dexamethasone group: 4 compared with 10 (P = 0.16, odds ratio 0.37, 95% confidence intervals 0.09-1.34). A subanalysis of mothers who delivered more than 24 hours after admission to the study and their infants revealed a significant reduction in perinatal deaths; 1 death in the dexamethasone group and 7 in the placebo group, P = 0.047 (Fisher's exact test). No woman in either group developed severe sepsis, and the incidence of sepsis in the women did not differ significantly. Eleven infants in each group developed sepsis. CONCLUSION: This is the first randomised trial in women with PPROM to compare the effects of the use of corticosteroids with placebo, where all women received prophylactic antibiotics concomitantly with the corticosteroids. A trend towards an improved perinatal outcome was demonstrated in the women who received dexamethasone. There was no increased risk of infection in the women or their infants where dexamethasone was administered. Administration of corticosteroids to women with PPROM has more advantages than disadvantages in developing countries.

Does successful completion of the Perinatal Education Programme result in improved obstetric practice?

OBJECTIVE: To determine whether successful completion of the Perinatal Education Programme (PEP) improves obstetric practice. METHOD: The three midwife obstetric units (MOUs) in a health district of Mpumalanga were included in the study. Two MOUs enrolled in the PEP and the third did not. A 'before-and-after' study design was used to assess any changes in practice, and to monitor whether any changes occurred in the district during the time of the study; data were also collected at the third MOU. Data were collected by scoring of the obstetric files after the patient had delivered. OUTCOME MEASURES: We ascertained whether the obstetric history, syphilis testing, blood group testing, haemoglobin measurement and uterine growth assessment were performed during antenatal care along with whether appropriate action was taken. For intrapartum care, estimation of fetal weight, the performance of pelvimetry, blood pressure monitoring, urine testing, evaluation of head above pelvis, fetal heart rate monitoring, monitoring of contractions and plotting of cervical dilatation, and whether the appropriate actions were taken, were assessed. RESULTS: Eight of the 13 midwives at the two MOUs completed the PEP and all demonstrated an improvement in knowledge. Case notes of 303 patients from the various clinics were studied. There was no change in the referral patterns of any of the clinics during the study period. The obstetric history was well documented, but in no group was there a satisfactory response to a detected problem; appropriate action was taken in between 0% and 12% of cases. Syphilis testing was performed in 56-82% of cases, with no difference between the groups. The haemoglobin level was measured in only 4-15% of patients, with no difference before or after completion of the PEP. Where a problem in uterine growth was detected, an appropriate response occurred in 0-8% of patients and no difference before or after completion of the PEP was ascertained. In all groups, estimation of fetal weight and pelvimetry were seldom performed, the urine and fetal heart rate documentation were moderately well done and the blood pressure monitoring, assessment of head above pelvis, monitoring of contractions and plotting of cervical dilatation were usually performed. No differences before or after the PEP were detected. Where problems were detected, appropriate actions taken during labour improved, but not significantly. CONCLUSION: Completion of the obstetric manual of the PEP improved the knowledge of the midwives but no alteration in practice was detected.

Low dose dopamine in postpartum pre-eclamptic women with oliguria: a double-blind, placebo controlled, randomised trial.

OBJECTIVE: To assess the effect of low dose dopamine on the urine output in postpartum pre-eclamptic or eclamptic women with oliguria. DESIGN: A double blind, randomised controlled study. SETTING: The high care area of the labour ward in a teaching hospital. SAMPLE: Forty postpartum pre-eclamptic women with oliguria, defined as < 30 mL/hour, who have not responded to a 300 mL crystalloid fluid challenge. INTERVENTION: Dopamine was infused at a rate of 1 to 5 microg/kg per minute, or sterile water was given as placebo in the same dilution. MAIN OUTCOME MEASURE: Urine output, blood pressure and pulse was measured for six hours before and for six hours after the intervention. RESULTS: Women who received dopamine (344 mL over 6 hours) showed a clinically and statistically significant (P = 0.0014, Mann-Whitney U test) higher median urine output compared with those receiving placebo (135 mL over 6 hours) for the duration of therapy. The respective 95% confidence intervals were 212.3 to 712.7 mL compared with 73.8 to 244.7 mL. No differences in blood pressure or pulse were found between the two groups. CONCLUSIONS: The use of low dose dopamine in a labour setting improved urine output in postpartum pre-eclamptic women with oliguria who had not responded to a single fluid challenge without a detrimental effect on the blood pressure or pulse.

The potential for preventing the delivery and perinatal mortality of low-birth-weight babies in a black urban population.

OBJECTIVE: To determine the potential for preventing the delivery and perinatal mortality of low-birth-weight (LBW) babies in a black urban population. DESIGN: Cross-sectional descriptive study. SETTING: All women delivering babies weighing less than 2,500 g at Kalafong Hospital in a 6-month period (December 1991-May 1992). MAIN OUTCOME MEASURES: The primary obstetric reason for delivery; whether the labour was of spontaneous onset or iatrogenic; whether labour was theoretically preventable using currently accepted practice; the number of patients in whom suppression of delivery was attempted in the theoretically preventable group; and the perinatal mortality rate of that group. RESULTS: There were 124 perinatal deaths (22.5%) in the 550 LBW babies delivered from 465 singleton pregnancies, 42 twin pregnancies and 1 triplet pregnancy. The primary obstetric reasons for delivery were spontaneous preterm labour (28%), hypertensive diseases (19%), premature rupture of membranes (18%), spontaneous labour in light-for-gestational-age babies (16%), unexplained intra-uterine deaths (8%), antepartum haemorrhage (8%) and other causes (3%). A medical decision to terminate the pregnancy before labour was made in 177 (34.8%) cases, the major reason being hypertensive diseases (84 mothers; 47.5%). In the remaining 331 mothers with spontaneous onset of labour, labour was theoretically preventable in 63 (19%) and prevention was only attempted in 12 (2.4% of the total mothers). The major reason for not attempting to suppress labour in the others was that the patients arrived too late at the hospital for intervention to take place. CONCLUSION: Hospital staff can do little to prevent the delivery of LBW babies in a black urban population.

The value of incorporating avoidable factors into perinatal audits.

OBJECTIVE: To assess whether incorporating a system of identifying, classifying and grading avoidable factors into a perinatal audit can be useful in identifying problem areas. DESIGN: Descriptive study. SETTING: Black urban population, Pretoria, South Africa. SUBJECTS: All perinatal deaths of infants weighing more than 1,000 g from urban areas served by Kalafong Hospital between August 1991 and July 1992. METHODS: All perinatal deaths were classified according to the primary obstetric cause of death and neonatal cause of death, and whether any avoidable factors were present which could have contributed to the death. RESULTS: The perinatal mortality rate was 26/1,000 deliveries. Avoidable factors occurred in 58% of perinatal deaths. Our problem areas which were immediately remedial were identified as labour management-related problems, administrative problems in obtaining syphilis results, and estimation of fetal weight. Other problem areas which need to be solved are patient education, early attendance at clinics, improved documentation and continuing education of medical personnel. CONCLUSION: The use of this classification of avoidable factors has enabled the detection of problem areas that can be improved immediately at very little cost.

Is ward evacuation for uncomplicated incomplete abortion under systemic analgesia safe and effective? A randomised clinical trial.

OBJECTIVE: To compare evacuation under systemic analgesia (fentanyl and midazolam) in a treatment room (ward group) with evacuation under general anaesthesia in theatre. DESIGN: A prospective randomised clinical trial. SETTING: A tertiary medical centre serving a black urban population. SUBJECTS: One hundred and forty-two patients with uncomplicated incomplete abortions. INTERVENTION: Randomisation into two groups, those for evacuation under systemic analgesia and those for evacuation under general anaesthesia. MAIN OUTCOME MEASURES: Both groups were compared in terms of safety, efficacy, acceptability, blood consumption and time delay between admission and evacuation. RESULTS: Significantly less blood was used in the ward group (37 units for 13 patients) than in the theatre group (65 units for 24 patients) (P < 0.03). Significantly less time was taken between admission and evacuation in the ward group (median 7 hours 15 minutes) than in the theatre group (median 12 hours 38 minutes) (P < 0.0003). Evacuation under fentanyl and midazolam was safe, effective and acceptable for the majority of patients compared with evacuation under general anaesthesia. CONCLUSION: Patients with uncomplicated incomplete abortions (uterine size equivalent to a pregnancy of 14 weeks' duration or less) can undergo evacuation safely and effectively under fentanyl and midazolam and have a significantly smaller chance of requiring a blood transfusion.