- Email: [email protected]
Expectant management of placenta previa: Cost-benefit analysis of outpatient treatment
May 1994 Am J Obstet Gynecol
Droste and Keil
supine. Because the fetal spine is lateral in most instances, 1 the coronal plane is imaged more conveniently than is the sagittal plane. Hence the coronal area formula, which has a similar 95% confidence limit, will be of greater use, because it is easier to measure ultrasonographically. Although this study does not directly address the impact of fetal bladder volume measurement errors on published estimates of hourly fetal urinary flow rates, the validity of these studies ' -'o should nevertheless be reevaluated with the corrected ovoid volume formula determined in this study. REFERENCES
1. Campbell S, Wladimiroff Jw, Dewhurst C]. 'The antenatal
measurement of fetal urine production. J Obstet Gynaecol Br Commonw 1973;80:680-6. 2. Rabinowitz R, Peters MT, Vyas S, Campbell S, Nicolaides KH. Measurement of fetal urine production in normal pregnancy by real-time ultrasonography. AM J OBSTET GYNECOL 1989;161:1264-6.
3. Chamberlain PF, Manning FA, Morrison I, Lange IR. Circadian rhythm in bladder volumes in the term human fetus. Obstet Gynecol 1984;64:657-60. 4. Wladimiroff JW, Campbell S. Fetal urine-production rates in normal and complicated pregnancy. Lancet 1974;1: 151-4. 5. Kurjak A, Kirkinen P, Latin V, Ivankovic D. Ultrasonic assessment of fetal kidney function in normal and complicated pregnancies. AM J OBSTET GYNECOL 1981; 141 :266-70. 6. Nicolaides KH, Peters MT, Vyas S, Rabinowitz R, Rosen DJD, Campbell S. Relation of rate of urine production to oxygen tension in small-for-gestational-age fetuses. AM J OBSTET GYNECOL 1990;162:387-91. 7. Zimmer EZ, Chao CR, Guy GP, Marks F, Fifer WP. Vibroacoustic stimulation evokes human fetal micturition. Obstet GynecoI1993;81:178-80. 8. Watson Wj, Katz VL, Seeds JW. Fetal urine output does not influence residual amniotic fluid volume after premature rupture of membranes. AM J OBSTET GYNECOL 1991; 164:64-5. 9. Kirshon B, Mari G, Moise Kj Jr. Indomethacin therapy in the treatment of symptomatic polyhydramnios. Obstet Gynecol 1990;75:202-5. 10. Kirshon B, Moise Kj Jr, Mari G, Willis R. Long-term indomethacin therapy decreases fetal urine output and results in oligohydramnios. Am J Perinat 1991 ;8:86-8.
Expectant management of placenta previa: Cost-benefit analysis of outpatient treatment Sabine Droste, MD, and Karen Keil, MD Madison, Wisconsin OBJECTIVE: In this study outpatient and inpatient expectant management for complete placenta previa were compared in terms of maternal and neonatal outcome and overall cost. STUDY DESIGN: We reviewed the outcomes and hospital costs of 72 mother-infant pairs where the pregnancy was complicated by second- or third-trimester placenta previa and was managed expectantly either with hospitalization or outpatient bed rest. The data were analyzed with the two-sided unpaired t test, x2 , and simple correlation analYSis. RESULTS: There were no differences in maternal morbidity as measured by estimated total blood loss, number of blood transfusions, nadir hematocrit, or need for emergency delivery. Fetal mortality was comparable in both groups, and there were no significant differences in neonatal morbidity as measured by gestational age, birth weight, 5-minute Apgar score, or occurrence of fetal distress. Among outpatients the number of maternal hospital days was reduced by 50% (p < 0.01). Outpatient management aChieved a hospital cost reduction of 48.5% for mothers (p < 0.001) and 39.4% for mother-infant pairs (p < 0.05). CONCLUSION: In selected patients outpatient management of complete placenta previa can be cost-effective and safe. (AM J OSSTET GVNECOL 1994;170:1254-7.)
Key words: Placenta previa, cost-benefit analysis, outpatient management
From the Dzvision of Maternal-Fetal Medicine, Department of Obstetncs and Gynecology, University of Wzsconsin-Madzson. Recezved for publicatIOn August 24, 1993; revISed November 10, 1993; accepted December 1, 1993. Reprint requests: Sabine Droste, MD, The Center for Perinatal Care/Maternal-Fetal Medicine, Menter Hospital-Park, 6-C, 202 S. Park St., Madison, WI 53715. Copynght © 1994 by Mosby-Year Book, Inc. 0002-9378/94 $3.00 + 0 6!1153326
1254
Over the past 70 years the management of placenta previa has changed from aggressive early delivery to a more expectant treatment approach with observation, transfusion, and tocolyis. On the basis of earlier observations of the medical effectiveness and cost-effectiveness of inpatient management of the patient with placenta previa, hospitalization for bed rest and aggressive
Droste and Keil
Volume 170, Number 5, Part I Am J Obstet Gynecol
expectant care have been advocated since the early 1980s. I Since then, however, the cost of hospitalization has steadily risen, and obstetricians are being asked to justify hospital admission to third-party payers. Enhanced scrutiny of the indications for inpatient management can be expected in the wake of anticipated health care reforms. The climate thus seems appropriate to reexamine the need for inpatient management of patients with placenta previa. This study is a retrospective cost-benefit analysis in a community hospital setting that compares outpatient expectant management with primary hospitalization in the management of second- and third-trimester complete placenta previa. Material and methods
The charts of all patients with the diagnosis of complete placenta previa were reviewed. Only patients who had a viable singleton gestation were included. All patients were cared for between January 1985 and December 1990 by community obstetricians at two midwestern hospitals with tertiary nursery facilities. In all cases the diagnosis was established by ultrasonographic localization of the placenta over the internal cervical os after 18 but before 36 weeks' gestation. Intraoperative confirmation of placenta previa at cesarean section was required for inclusion into the study. Patients with bleeding heavy enough to necessitate delivery within < 48 hours of initial diagnosis were not included. The patients were stratified into two groups on the basis of their physician'S intent to treat on an inpatient or an outpatient basis. The inpatient group was hospitalized from the first bleeding episode onward or was admitted electively on reaching 24 weeks' gestation. The outpatient group was observed on home bed rest and pelvic rest. These patients were hospitalized for bleeding episodes if it was deemed necessary by the attending obstetrician. In both groups transfusions, steroids, tocolytics (magnesium sulfate or terbutaline), and antepartum testing were used variably at the managing physician's discretion. All patients were delivered by cesarean section, either on an emergency basis for maternal bleeding or fetal distress or electively after fetal pulmonary maturity had been demonstrated by amniocentesis. The maternal medical records were abstracted for demographic characteristics, obstetric history, gestational age at diagnosis, gestational age at the time of the first bleeding episode, total number of bleeding episodes, estimated total blood loss, nadir hematocrit, number of blood transfusions, number of hospital admissions, total number of inpatient days, and the need for emergency delivery. The corresponding neonatal records were reviewed and abstracted for gestational
1255
age at birth as determined by best obstetric dates confirmed by Dubowitz score, birth weight, presence or absence of fetal distress as indicated by the operative note, 5-minute Apgar score, and total nursery days. Total hospital charges for each mother-infant pair were obtained from the hospitals' billing offices. These charges did not include physicians' bills. The data were analyzed by means of a microcomputer statistical software package (StatView 512, Abacus Concepts, 1986) with the two-sided unpaired t test and simple correlation and X2 analysis. A p value of s; 0.05 was considered statistically significant. Results
Seventy-two patients with complete placenta previa were included in the analysis. Thirty-five patients were managed as inpatients and 37 as outpatients. Assignment into the inpatient versus outpatient treatment group was based on the managing physician's documented intent at the time of diagnosis. Maternal demographic data are shown in Table I. There were no significant differences in age, gravidity, parity, or number of prior cesarean sections. Gestational ages at the time of the first bleeding episode were comparable, but there was a difference in gestational age at initial diagnosis. Outpatients were diagnosed approximately 3 weeks earlier, at 26.9 ± 5.0 weeks, whereas the gestational age at diagnosis was 30.2 ± 4.3 weeks in the inpatient management group (p < 0.01). Four of 35 patients in the inpatient and six of 37 patients in the outpatient groups experienced initial bleeding episodes before 24 weeks' gestation. The earlier gestational age at diagnosis was significantly correlated with assignment to the outpatient management group (R2 = 0.11, P < 0.01). Maternal and neonatal morbidity are summarized in Table II. Estimated total blood loss, number of bleeding episodes, gestational age at delivery, birth weight, and the incidence of 5-minute Apgar scores < 7 were all comparable between the two groups. Thirteen patients in each group were delivered by emergency cesarean section; the remainder were delivered electively. There was one neonatal death in each group. Among the inpatients one mother experienced bleeding at 18 weeks' gestation. She was electively hospitalized at 23 weeks and 6 days later was delivered of a 610 gm fetus by emergency cesarean section for intractable hemorrhage. The newborn died on the fourth day of life from complications of extreme prematurity. One patient in the outpatient treatment group was also diagnosed at 18 weeks' gestation; she had multiple bleeding episodes at home until 23 weeks, when she was briefly hospitalized for bleeding and discharged again. A massive hemorrhage and fetal distress necessitated emergency cesarean delivery 3.5 weeks later. Complications of
May 1994
1256 Droste and Keil
Am J Obstet Gynecol
Table I. Demographic data Inpatzent 29.1 1.5 0.2 30.2 29.6
Age (yr) Parity No. of prior cesarean sections GestatIonal age at diagnosis (wk) Gestational age at first bleeding (wk)
± 5.0 ± 1.7 ± 0.5
± 4.3 ± 4.8
Outpatzent 29.7 1.4 0.5 27.0 28.0
± 4.5 ± 1.2 ± 1.2 ± 5.0 ± 4.3
SIgnificance
NS NS NS
P<
0.01
NS
Values are mean ± 1 SD. NS, Not significant.
Table II. Outcome Inpatzent Maternal outcome Bleeding episodes Estimated blood loss (ml) Nadir hematocrit (%) Units of blood transfused Neonatal outcome GestatIonal age (wk) Birth weight (gm) 5 min Apgar score < 7
4.0 576.1 27.1 1.5
± ± ± ±
2.9 467.4 2.7 2.1
34.7 ± 3.1 2633.3 ± 747.7
2
Outpattent 4.1 538.2 26.9 1.7
± 3.5 ± 574.2 ± 3.7
± 2.2
35.1 ± 2.7 2695.8 ± 625.9 4
Szgnificance
NS NS NS NS NS NS NS
Values are mean ± 1 SD. NS, Not significant.
severe respiratory distress syndrome resulted in death of the neonate on day 2 of life. The cost analysis is presented in Table III. In the inpatient group all patients had a single hospital admission, except for one patient who had a bleeding episode at 25 weeks' gestation that was not further evaluated, but for which she was briefly hospitalized. The diagnosis of complete placenta previa in this patient was not established until 31 weeks' gestation, when she was admitted for inpatient observation. Patients in the outpatient treatment group had significantly more separate admissions (p < 0.01), whereas inpatients remained hospitalized longer overall (p < 0.01) with proportionally increased hospital cost. Conversely, there were no apparent differences in the duration of neonatal hospital stay nor in neonatal hospital cost between the two treatment groups. The reduction in overall hospital cost for mother-infant pairs in the outpatient group was statistically significant (p < 0.05). Comment
The management of complete placenta previa has undergone considerable evolution over time. Until the 1930s vaginal delivery, often by means of mechanical cervical dilating devices, was the rule, but it was accompanied by unacceptably high maternal morbidity and mortality. In 1927 BilF advocated the use of prophylactic blood transfusions and cesarean section in an effort to reduce maternal mortality. In 1945 Williamson and Greely' proposed an expectant management strategy with avoidance of digital examinations and prolon-
gation of gestation with maternal hospitalization and transfusion support, followed by cesarean delivery. Although this approach rapidly became the standard of care, perinatal mortality remained high at approximately 20% to 30%'<' , A significant reduction in perinatal mortality was not achieved until treatment was individualized and tocolysis, antepartum fetal surveillance, and timed elective delivery after biochemical determination of pulmonary maturity were added to the therapeutic regimen. 6 This "aggressive expectant" treatment protocol has been thought to require inpatient observation. In 1984 D'Angelo and Irwin I published a retrospective cost-benefit analysis that supported the overall cost-effectiveness of inpatient management. In their study of 38 cases of placenta previa any savings resulting from maternal outpatient expectant management were more than offset by a significant increase in neonatal morbidity, with reduced gestational ages and birth weights. This resulted in an overall hospital cost for maternal-neonatal pairs that was 69% higher in the group of patients managed as outpatients. In addition, there was a tendency toward greater transfusion requirements in patients managed with home bed rest. However, in these patients delivery was primarily prompted by severe hemorrhage rather than fetal pulmonary maturity, which could have biased this observation. On the basis of these findings the authors made the recommendation to hospitalize all patients with placenta previa when a gestational age of 28 weeks was achieved, which represented the then approximate minimum gestational age for neonatal survival.
Droste and Keil
Volume 170, Number 5, Part 1 Am J Obstet Gynecol
1257
Table III. Cost analysis Inpattent
Maternal admiSSIOns Maternal inpatient days Maternal hospital cost (U's. $) Neonatal inpatient days Neonatal hospital cost (U.S. $) Cost per mother-infant pair (U.S. $)
1.03 30.97 11,966.17 11.77 7,687.14 19,857.98
Outpatient
± 0.17
± 20.81
± 6,403.28 ± 15.80 ± 15,181.71 ± 15,119.54
1.73 15.46 6,154.03 12.42 6,776.51 12,038.64
SIgnificance
± 0.65
p < 0.01
± 4,240.61
P < 0.01
± 11.78
± 10.95 ± 9,242.28 ± 10,426.72
P<
0.01
NS NS P < 0.05
Values are mean ± 1 SD. NS, Not significant.
Since the publication of D'Angelo and Irwin, significant advances in neonatal care have pushed back the age for fetal viability to 24 to 25 weeks' gestation. 7 An inpatient expectant management approach in the 1990s would thus add 3 to 4 weeks of maternal hospitalization at considerable additional expense. Our study fails to confirm the cost-effectiveness of inpatient management for complete placenta previa. In contrast to the observations of D'Angelo and Irwin, savings incurred by maternal outpatient treatment were not offset by additional costs attributable to increases in neonatal morbidity. Gestational ages at birth, birth weights, duration of nursery stay, and neonatal hospital charges were all comparable between the two treatment groups, and maternal hospital costs were markedly reduced by outpatient management. This resulted in an average 39.9% reduction of hospital cost for motherinfant pairs, for a cost savings of over $7500 per pregnancy. Although these data seem persuasive, the retrospective design of our study does have significant limitations. Because patients were managed by different obstetricians at two hospitals, protocols for antenatal surveillance and use of antepartum steroids varied. The use of aggressive tocolysis for bleeding episodes appeared to increase over the duration of the study period as magnesium sulfate gained popularity as a tocolytic. The potential confounding effects of these variables could not be adequately examined. Furthermore, patient selection for inpatient versus outpatient management was almost surely biased. More aggressive hospitalization might have been chosen for less compliant patients in lower socioeconomic classes with less domestic support who resided at a farther distance from the hospital. This possible confounder is suggested by the somewhat later gestational age at diagnosis in the group of patients managed on an inpatient basis. The effects of differential patient selection could have biased the outcomes in favor of what can be assumed to be the more compliant outpatient group. This might have obscured outcome differences between the two treatment approaches. Conversely, patients who first bled at an earlier gestational age were, probably because of cost considerations, more likely to have been selected for
outpatient observation, which might have biased the outcome in the opposite direction. Only a prospective, randomized study design with uniform management protocols could address these issues adequately. Realistically, however, careful and somewhat subjective patient selection is likely to be a requirement for the safe outpatient management of placenta previa. Minority women, who are presumably also economically disadvantaged, have been shown to have a significantly greater risk from placenta previa. 8 Patients who have economic disadvantages, little domestic support, young children at home, or transportation difficulties are thus unlikely to be good candidates for this treatment approach. Maternal hospitalization for recurrent bleeding episodes, to provide tocolysis and transfusion, was liberally used to manage the outpatients in our study. Nonetheless, maternal inpatient days were reduced by 50%. No evidence of detrimental effects on maternal or neonatal health could be demonstrated from this approach. We thus conclude that the judicious outpatient management of carefully selected patients with complete placenta previa can be both cost-effective and safe. REFERENCES
1. D'Angelo LJ, Irwin LF. Conservative management of placenta previa: a cost-benefit analysis. A\1 J OBSTET GY'JECOL 1984;149:320-6. 2. Bill AB. The treatment of placenta previa by prophylactic blood transfusion and cesarean section. AAI J OBS'IEI G\NECOL 1927;14:523-9. 3. Williamson HC, Greely AV. Management of placenta previa: a twelve-year study. AM J OBSIET GYNECOL 1945;50: 398-406. 4. Brenner WE, Edelman DA, Hendricks CH. Characteristics of patients with placenta previa and results of "expectant management." A1\IJ OBSTET GYNECOL 1978;132:180-91. 5. Crenshaw C, Jones DED, Parker RT. Placenta previa: a survey of twenty years experience with improved perinatal survival by expectant therapy and cesarean delivery. Obstet Gynecol Surv 1973;28:461-70. 6. Cotton DB, Read JA, Paul RA, Quilligan EJ. The conservative aggressive management of placenta previa. AM J OB&TET GYNECOL 1980;137:687-95. 7. Phelps DL, Brown DR, Tung B, et al. 28 day survival rates of 6676 neonates with birthweights of 1250 grams or le%. Pediatrics 1991 ;87:7 -17. 8. Iyasu S, Saftlas AK, Rowley DL, et al. The epidemiology of placenta previa in the United States, 1979 through 1987. AM J OBSIET GYNECOL 1993; 168: 1424-9.
Droste and Keil
supine. Because the fetal spine is lateral in most instances, 1 the coronal plane is imaged more conveniently than is the sagittal plane. Hence the coronal area formula, which has a similar 95% confidence limit, will be of greater use, because it is easier to measure ultrasonographically. Although this study does not directly address the impact of fetal bladder volume measurement errors on published estimates of hourly fetal urinary flow rates, the validity of these studies ' -'o should nevertheless be reevaluated with the corrected ovoid volume formula determined in this study. REFERENCES
1. Campbell S, Wladimiroff Jw, Dewhurst C]. 'The antenatal
measurement of fetal urine production. J Obstet Gynaecol Br Commonw 1973;80:680-6. 2. Rabinowitz R, Peters MT, Vyas S, Campbell S, Nicolaides KH. Measurement of fetal urine production in normal pregnancy by real-time ultrasonography. AM J OBSTET GYNECOL 1989;161:1264-6.
3. Chamberlain PF, Manning FA, Morrison I, Lange IR. Circadian rhythm in bladder volumes in the term human fetus. Obstet Gynecol 1984;64:657-60. 4. Wladimiroff JW, Campbell S. Fetal urine-production rates in normal and complicated pregnancy. Lancet 1974;1: 151-4. 5. Kurjak A, Kirkinen P, Latin V, Ivankovic D. Ultrasonic assessment of fetal kidney function in normal and complicated pregnancies. AM J OBSTET GYNECOL 1981; 141 :266-70. 6. Nicolaides KH, Peters MT, Vyas S, Rabinowitz R, Rosen DJD, Campbell S. Relation of rate of urine production to oxygen tension in small-for-gestational-age fetuses. AM J OBSTET GYNECOL 1990;162:387-91. 7. Zimmer EZ, Chao CR, Guy GP, Marks F, Fifer WP. Vibroacoustic stimulation evokes human fetal micturition. Obstet GynecoI1993;81:178-80. 8. Watson Wj, Katz VL, Seeds JW. Fetal urine output does not influence residual amniotic fluid volume after premature rupture of membranes. AM J OBSTET GYNECOL 1991; 164:64-5. 9. Kirshon B, Mari G, Moise Kj Jr. Indomethacin therapy in the treatment of symptomatic polyhydramnios. Obstet Gynecol 1990;75:202-5. 10. Kirshon B, Moise Kj Jr, Mari G, Willis R. Long-term indomethacin therapy decreases fetal urine output and results in oligohydramnios. Am J Perinat 1991 ;8:86-8.
Expectant management of placenta previa: Cost-benefit analysis of outpatient treatment Sabine Droste, MD, and Karen Keil, MD Madison, Wisconsin OBJECTIVE: In this study outpatient and inpatient expectant management for complete placenta previa were compared in terms of maternal and neonatal outcome and overall cost. STUDY DESIGN: We reviewed the outcomes and hospital costs of 72 mother-infant pairs where the pregnancy was complicated by second- or third-trimester placenta previa and was managed expectantly either with hospitalization or outpatient bed rest. The data were analyzed with the two-sided unpaired t test, x2 , and simple correlation analYSis. RESULTS: There were no differences in maternal morbidity as measured by estimated total blood loss, number of blood transfusions, nadir hematocrit, or need for emergency delivery. Fetal mortality was comparable in both groups, and there were no significant differences in neonatal morbidity as measured by gestational age, birth weight, 5-minute Apgar score, or occurrence of fetal distress. Among outpatients the number of maternal hospital days was reduced by 50% (p < 0.01). Outpatient management aChieved a hospital cost reduction of 48.5% for mothers (p < 0.001) and 39.4% for mother-infant pairs (p < 0.05). CONCLUSION: In selected patients outpatient management of complete placenta previa can be cost-effective and safe. (AM J OSSTET GVNECOL 1994;170:1254-7.)
Key words: Placenta previa, cost-benefit analysis, outpatient management
From the Dzvision of Maternal-Fetal Medicine, Department of Obstetncs and Gynecology, University of Wzsconsin-Madzson. Recezved for publicatIOn August 24, 1993; revISed November 10, 1993; accepted December 1, 1993. Reprint requests: Sabine Droste, MD, The Center for Perinatal Care/Maternal-Fetal Medicine, Menter Hospital-Park, 6-C, 202 S. Park St., Madison, WI 53715. Copynght © 1994 by Mosby-Year Book, Inc. 0002-9378/94 $3.00 + 0 6!1153326
1254
Over the past 70 years the management of placenta previa has changed from aggressive early delivery to a more expectant treatment approach with observation, transfusion, and tocolyis. On the basis of earlier observations of the medical effectiveness and cost-effectiveness of inpatient management of the patient with placenta previa, hospitalization for bed rest and aggressive
Droste and Keil
Volume 170, Number 5, Part I Am J Obstet Gynecol
expectant care have been advocated since the early 1980s. I Since then, however, the cost of hospitalization has steadily risen, and obstetricians are being asked to justify hospital admission to third-party payers. Enhanced scrutiny of the indications for inpatient management can be expected in the wake of anticipated health care reforms. The climate thus seems appropriate to reexamine the need for inpatient management of patients with placenta previa. This study is a retrospective cost-benefit analysis in a community hospital setting that compares outpatient expectant management with primary hospitalization in the management of second- and third-trimester complete placenta previa. Material and methods
The charts of all patients with the diagnosis of complete placenta previa were reviewed. Only patients who had a viable singleton gestation were included. All patients were cared for between January 1985 and December 1990 by community obstetricians at two midwestern hospitals with tertiary nursery facilities. In all cases the diagnosis was established by ultrasonographic localization of the placenta over the internal cervical os after 18 but before 36 weeks' gestation. Intraoperative confirmation of placenta previa at cesarean section was required for inclusion into the study. Patients with bleeding heavy enough to necessitate delivery within < 48 hours of initial diagnosis were not included. The patients were stratified into two groups on the basis of their physician'S intent to treat on an inpatient or an outpatient basis. The inpatient group was hospitalized from the first bleeding episode onward or was admitted electively on reaching 24 weeks' gestation. The outpatient group was observed on home bed rest and pelvic rest. These patients were hospitalized for bleeding episodes if it was deemed necessary by the attending obstetrician. In both groups transfusions, steroids, tocolytics (magnesium sulfate or terbutaline), and antepartum testing were used variably at the managing physician's discretion. All patients were delivered by cesarean section, either on an emergency basis for maternal bleeding or fetal distress or electively after fetal pulmonary maturity had been demonstrated by amniocentesis. The maternal medical records were abstracted for demographic characteristics, obstetric history, gestational age at diagnosis, gestational age at the time of the first bleeding episode, total number of bleeding episodes, estimated total blood loss, nadir hematocrit, number of blood transfusions, number of hospital admissions, total number of inpatient days, and the need for emergency delivery. The corresponding neonatal records were reviewed and abstracted for gestational
1255
age at birth as determined by best obstetric dates confirmed by Dubowitz score, birth weight, presence or absence of fetal distress as indicated by the operative note, 5-minute Apgar score, and total nursery days. Total hospital charges for each mother-infant pair were obtained from the hospitals' billing offices. These charges did not include physicians' bills. The data were analyzed by means of a microcomputer statistical software package (StatView 512, Abacus Concepts, 1986) with the two-sided unpaired t test and simple correlation and X2 analysis. A p value of s; 0.05 was considered statistically significant. Results
Seventy-two patients with complete placenta previa were included in the analysis. Thirty-five patients were managed as inpatients and 37 as outpatients. Assignment into the inpatient versus outpatient treatment group was based on the managing physician's documented intent at the time of diagnosis. Maternal demographic data are shown in Table I. There were no significant differences in age, gravidity, parity, or number of prior cesarean sections. Gestational ages at the time of the first bleeding episode were comparable, but there was a difference in gestational age at initial diagnosis. Outpatients were diagnosed approximately 3 weeks earlier, at 26.9 ± 5.0 weeks, whereas the gestational age at diagnosis was 30.2 ± 4.3 weeks in the inpatient management group (p < 0.01). Four of 35 patients in the inpatient and six of 37 patients in the outpatient groups experienced initial bleeding episodes before 24 weeks' gestation. The earlier gestational age at diagnosis was significantly correlated with assignment to the outpatient management group (R2 = 0.11, P < 0.01). Maternal and neonatal morbidity are summarized in Table II. Estimated total blood loss, number of bleeding episodes, gestational age at delivery, birth weight, and the incidence of 5-minute Apgar scores < 7 were all comparable between the two groups. Thirteen patients in each group were delivered by emergency cesarean section; the remainder were delivered electively. There was one neonatal death in each group. Among the inpatients one mother experienced bleeding at 18 weeks' gestation. She was electively hospitalized at 23 weeks and 6 days later was delivered of a 610 gm fetus by emergency cesarean section for intractable hemorrhage. The newborn died on the fourth day of life from complications of extreme prematurity. One patient in the outpatient treatment group was also diagnosed at 18 weeks' gestation; she had multiple bleeding episodes at home until 23 weeks, when she was briefly hospitalized for bleeding and discharged again. A massive hemorrhage and fetal distress necessitated emergency cesarean delivery 3.5 weeks later. Complications of
May 1994
1256 Droste and Keil
Am J Obstet Gynecol
Table I. Demographic data Inpatzent 29.1 1.5 0.2 30.2 29.6
Age (yr) Parity No. of prior cesarean sections GestatIonal age at diagnosis (wk) Gestational age at first bleeding (wk)
± 5.0 ± 1.7 ± 0.5
± 4.3 ± 4.8
Outpatzent 29.7 1.4 0.5 27.0 28.0
± 4.5 ± 1.2 ± 1.2 ± 5.0 ± 4.3
SIgnificance
NS NS NS
P<
0.01
NS
Values are mean ± 1 SD. NS, Not significant.
Table II. Outcome Inpatzent Maternal outcome Bleeding episodes Estimated blood loss (ml) Nadir hematocrit (%) Units of blood transfused Neonatal outcome GestatIonal age (wk) Birth weight (gm) 5 min Apgar score < 7
4.0 576.1 27.1 1.5
± ± ± ±
2.9 467.4 2.7 2.1
34.7 ± 3.1 2633.3 ± 747.7
2
Outpattent 4.1 538.2 26.9 1.7
± 3.5 ± 574.2 ± 3.7
± 2.2
35.1 ± 2.7 2695.8 ± 625.9 4
Szgnificance
NS NS NS NS NS NS NS
Values are mean ± 1 SD. NS, Not significant.
severe respiratory distress syndrome resulted in death of the neonate on day 2 of life. The cost analysis is presented in Table III. In the inpatient group all patients had a single hospital admission, except for one patient who had a bleeding episode at 25 weeks' gestation that was not further evaluated, but for which she was briefly hospitalized. The diagnosis of complete placenta previa in this patient was not established until 31 weeks' gestation, when she was admitted for inpatient observation. Patients in the outpatient treatment group had significantly more separate admissions (p < 0.01), whereas inpatients remained hospitalized longer overall (p < 0.01) with proportionally increased hospital cost. Conversely, there were no apparent differences in the duration of neonatal hospital stay nor in neonatal hospital cost between the two treatment groups. The reduction in overall hospital cost for mother-infant pairs in the outpatient group was statistically significant (p < 0.05). Comment
The management of complete placenta previa has undergone considerable evolution over time. Until the 1930s vaginal delivery, often by means of mechanical cervical dilating devices, was the rule, but it was accompanied by unacceptably high maternal morbidity and mortality. In 1927 BilF advocated the use of prophylactic blood transfusions and cesarean section in an effort to reduce maternal mortality. In 1945 Williamson and Greely' proposed an expectant management strategy with avoidance of digital examinations and prolon-
gation of gestation with maternal hospitalization and transfusion support, followed by cesarean delivery. Although this approach rapidly became the standard of care, perinatal mortality remained high at approximately 20% to 30%'<' , A significant reduction in perinatal mortality was not achieved until treatment was individualized and tocolysis, antepartum fetal surveillance, and timed elective delivery after biochemical determination of pulmonary maturity were added to the therapeutic regimen. 6 This "aggressive expectant" treatment protocol has been thought to require inpatient observation. In 1984 D'Angelo and Irwin I published a retrospective cost-benefit analysis that supported the overall cost-effectiveness of inpatient management. In their study of 38 cases of placenta previa any savings resulting from maternal outpatient expectant management were more than offset by a significant increase in neonatal morbidity, with reduced gestational ages and birth weights. This resulted in an overall hospital cost for maternal-neonatal pairs that was 69% higher in the group of patients managed as outpatients. In addition, there was a tendency toward greater transfusion requirements in patients managed with home bed rest. However, in these patients delivery was primarily prompted by severe hemorrhage rather than fetal pulmonary maturity, which could have biased this observation. On the basis of these findings the authors made the recommendation to hospitalize all patients with placenta previa when a gestational age of 28 weeks was achieved, which represented the then approximate minimum gestational age for neonatal survival.
Droste and Keil
Volume 170, Number 5, Part 1 Am J Obstet Gynecol
1257
Table III. Cost analysis Inpattent
Maternal admiSSIOns Maternal inpatient days Maternal hospital cost (U's. $) Neonatal inpatient days Neonatal hospital cost (U.S. $) Cost per mother-infant pair (U.S. $)
1.03 30.97 11,966.17 11.77 7,687.14 19,857.98
Outpatient
± 0.17
± 20.81
± 6,403.28 ± 15.80 ± 15,181.71 ± 15,119.54
1.73 15.46 6,154.03 12.42 6,776.51 12,038.64
SIgnificance
± 0.65
p < 0.01
± 4,240.61
P < 0.01
± 11.78
± 10.95 ± 9,242.28 ± 10,426.72
P<
0.01
NS NS P < 0.05
Values are mean ± 1 SD. NS, Not significant.
Since the publication of D'Angelo and Irwin, significant advances in neonatal care have pushed back the age for fetal viability to 24 to 25 weeks' gestation. 7 An inpatient expectant management approach in the 1990s would thus add 3 to 4 weeks of maternal hospitalization at considerable additional expense. Our study fails to confirm the cost-effectiveness of inpatient management for complete placenta previa. In contrast to the observations of D'Angelo and Irwin, savings incurred by maternal outpatient treatment were not offset by additional costs attributable to increases in neonatal morbidity. Gestational ages at birth, birth weights, duration of nursery stay, and neonatal hospital charges were all comparable between the two treatment groups, and maternal hospital costs were markedly reduced by outpatient management. This resulted in an average 39.9% reduction of hospital cost for motherinfant pairs, for a cost savings of over $7500 per pregnancy. Although these data seem persuasive, the retrospective design of our study does have significant limitations. Because patients were managed by different obstetricians at two hospitals, protocols for antenatal surveillance and use of antepartum steroids varied. The use of aggressive tocolysis for bleeding episodes appeared to increase over the duration of the study period as magnesium sulfate gained popularity as a tocolytic. The potential confounding effects of these variables could not be adequately examined. Furthermore, patient selection for inpatient versus outpatient management was almost surely biased. More aggressive hospitalization might have been chosen for less compliant patients in lower socioeconomic classes with less domestic support who resided at a farther distance from the hospital. This possible confounder is suggested by the somewhat later gestational age at diagnosis in the group of patients managed on an inpatient basis. The effects of differential patient selection could have biased the outcomes in favor of what can be assumed to be the more compliant outpatient group. This might have obscured outcome differences between the two treatment approaches. Conversely, patients who first bled at an earlier gestational age were, probably because of cost considerations, more likely to have been selected for
outpatient observation, which might have biased the outcome in the opposite direction. Only a prospective, randomized study design with uniform management protocols could address these issues adequately. Realistically, however, careful and somewhat subjective patient selection is likely to be a requirement for the safe outpatient management of placenta previa. Minority women, who are presumably also economically disadvantaged, have been shown to have a significantly greater risk from placenta previa. 8 Patients who have economic disadvantages, little domestic support, young children at home, or transportation difficulties are thus unlikely to be good candidates for this treatment approach. Maternal hospitalization for recurrent bleeding episodes, to provide tocolysis and transfusion, was liberally used to manage the outpatients in our study. Nonetheless, maternal inpatient days were reduced by 50%. No evidence of detrimental effects on maternal or neonatal health could be demonstrated from this approach. We thus conclude that the judicious outpatient management of carefully selected patients with complete placenta previa can be both cost-effective and safe. REFERENCES
1. D'Angelo LJ, Irwin LF. Conservative management of placenta previa: a cost-benefit analysis. A\1 J OBSTET GY'JECOL 1984;149:320-6. 2. Bill AB. The treatment of placenta previa by prophylactic blood transfusion and cesarean section. AAI J OBS'IEI G\NECOL 1927;14:523-9. 3. Williamson HC, Greely AV. Management of placenta previa: a twelve-year study. AM J OBSIET GYNECOL 1945;50: 398-406. 4. Brenner WE, Edelman DA, Hendricks CH. Characteristics of patients with placenta previa and results of "expectant management." A1\IJ OBSTET GYNECOL 1978;132:180-91. 5. Crenshaw C, Jones DED, Parker RT. Placenta previa: a survey of twenty years experience with improved perinatal survival by expectant therapy and cesarean delivery. Obstet Gynecol Surv 1973;28:461-70. 6. Cotton DB, Read JA, Paul RA, Quilligan EJ. The conservative aggressive management of placenta previa. AM J OB&TET GYNECOL 1980;137:687-95. 7. Phelps DL, Brown DR, Tung B, et al. 28 day survival rates of 6676 neonates with birthweights of 1250 grams or le%. Pediatrics 1991 ;87:7 -17. 8. Iyasu S, Saftlas AK, Rowley DL, et al. The epidemiology of placenta previa in the United States, 1979 through 1987. AM J OBSIET GYNECOL 1993; 168: 1424-9.