Furosemide for prevention of morbidity in indomethacin-treated infants with patent ductus arteriosus

Brion LP, Campbell DE

Background - Methods - Results - Characteristics of Included Studies - References - Data Tables and Graphs

Dates

Date edited: 11/07/2007
Date of last substantive update: 09/04/2001
Date of last minor update: 23/04/2007
Date next stage expected 17/05/2009
Protocol first published:
Review first published: Issue 3, 2001

Contact reviewer

Luc P. Brion, M.D.
Professor of Pediatrics
Division of Neonatal-Perinatal Medicine
University of Texas Southwestern at Dallas
5323 Harry Hines Boulevard
Dallas
Texas USA
75390-9063
Telephone 1: 1 214 648 2060
Facsimile: 1 214 648 2481
E-mail: Luc.Brion@UTSouthwestern.edu

Contribution of reviewers

Internal sources of support

Div. of Neonatology, Albert Einstein College of Medicine and Montefiore Med Ctr, USA

External sources of support

Grant-in-Aid 9650445N, American Heart Association (LPB), USA

What's new

This review updates the existing review of "Furosemide for prevention of morbidity in indomethacin-treated infants with patent ductus arteriosus" published in The Cochrane Library, Issue 1, 2004 (Brion 2004).

For this update, a repeat literature search was conducted in April 2007. No new trials were identified. A synopsis was added to the review.

Dates

Date review re-formatted: 29/08/1999
Date new studies sought but none found: 23/04/2007
Date new studies found but not yet included/excluded: / /
Date new studies found and included/excluded: / /
Date reviewers' conclusions section amended: / /
Date comment/criticism added: / /
Date response to comment/criticisms added: / /

Text of review

Synopsis

There is no strong evidence of benefit from routine use furosemide, a loop diuretic, in preterm babies receiving indomethacin for treatment of patent ductus arteriosus.

A blood vessel (ductus arteriosus), which is required for blood circulation for the fetus in the womb, closes soon after birth in babies born around the expected date of delivery (term infants). Babies born early (preterm) may develop symptoms if they do not close that blood vessel after birth. Preterm infants who have symptoms due to the ductus arteriosus may receive therapy (indomethacin) for closing that vessel. Indomethacin may decrease kidney function and the amount of urine. Furosemide, a medication which reduces body water (diuretic), might help limit the effects of indomethacin on the kidney. This review analyzed the effects of furosemide on preterm babies receiving indomethacin to close the ductus arteriosus. The review of trials found not enough evidence to recommend routine use of furosemide in preterm infants who receive indomethacin for closing a ductus arteriosus.

Abstract

Background

Objectives

The primary objectives of this review were to assess (1) whether furosemide affects the incidence of failure of ductal closure after indomethacin and that of indomethacin-related toxicity and (2) the effect of furosemide on mid-term and long-term outcome. The secondary objective was to determine whether the effect of furosemide on renal function and water balance depends on prior extracellular volume (assessed by blood urea nitrogen [BUN]/creatinine ratio).

Search strategy

Electronic databases (MEDLINE, EMBASE and the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library) and selected abstract books, without language restriction were searched. For the latest update, database searches were done on April 1, 2007 and Issue 1, 2007 of The Cochrane Library.

Selection criteria

Studies with (1) random allocation to either indomethacin alone or indomethacin and furosemide and (2) analysis of either short-term risk-benefit ratio of furosemide, mid- or long-term outcome, or the relationship between extracellular volume at study entry and changes in renal function were selected.

Data collection & analysis

Studies were assessed for possible bias and for quality of assessment of ductal patency. Categorical variables were assessed using relative risk and absolute risk reduction. The effects of furosemide on renal function and fluid balance were assessed by comparing changes from baseline in the treatment group with those in controls. Subsets were determined a priori based on BUN/creatinine ratio at study entry.

Main results

All three studies fulfilling the entry criteria had limitations, including possible or definite bias. There was substantial heterogeneity among studies.
Furosemide administration did not significantly increase the risk of failure of ductal closure; however, sample size was insufficient to rule out even a 31% increase. In the subset with initial BUN/creatinine ratio > 20 mg/mg, two of 18 patients receiving furosemide could not complete a three-dose course of indomethacin because of toxicity. Minimal or no information was available about any of the other main outcome variables. Furosemide increased urine output regardless of the initial BUN/creatinine ratio, leading to a 5% weight loss during a three-dose course, an undesired effect in patients with initial BUN/creatinine ratio > 20 mg/mg. Furosemide increased creatinine clearance only in patients with initial BUN/creatinine ratio < 20 mg/mg.

Reviewers' conclusions

There is not enough evidence to support the administration of furosemide to premature infants treated with indomethacin for symptomatic patent ductus arteriosus. Furosemide appears to be contraindicated in the presence of dehydration in those infants.

Background

Inhibition of prostaglandin synthesis mediates closure of the ductus arteriosus and renal side effects after indomethacin administration. Because furosemide increases prostaglandin production, furosemide administration could help prevent indomethacin-related toxicity but also could affect ductal response to indomethacin. This systematic review was undertaken to assess whether current evidence justifies administering furosemide to preterm infants treated with indomethacin for symptomatic patent ductus arteriosus.

Side effects of indomethacin on renal function and fluid balance:
In premature infants with symptomatic patent ductus arteriosus, indomethacin administration may induce various renal and hydromineral side-effects, including proteinuria, oliguria, renal failure, hyperkalemia, and hyponatremia. Indomethacin-related transient renal dysfunction is associated with inhibition of prostaglandin synthesis, a reduction in renal blood flow (van Bel 1991), a decrease in plasma renin activity from the high levels that are attributed to renal hypoperfusion, and a transient rise in the plasma level of arginine vasopressin (Seyberth 1983). Even when administered at a low prophylactic dose, indomethacin may transiently impair renal function (Fowlie 1997).

High risk for indomethacin-induced nephrotoxicity in association with oliguria preceding indomethacin administration (Barrington 1994) has been reported, as well as with a high serum concentration of indomethacin (Friedman 1982), with concomitant administration of other nephrotoxic drugs and with persistent ductal patency despite therapy (Lin 1995). The risk for indomethacin-related nephrotoxicity may be decreased by slow or continuous infusion of indomethacin (Zecca 1994; Christmann 1998), by prior volume expansion (Leititis 1987) or by dopamine infusion (Seri 1984; Cochran 1989).

Interaction between furosemide, indomethacin and prostaglandins:
A randomized controlled trial comparing furosemide to chlorothiazide has shown that furosemide increases the incidence of patent ductus arteriosus in premature infants with respiratory distress syndrome. This occurs presumably through a prostaglandin-mediated mechanism (Green 1983). Furthermore, furosemide has been shown to increase urinary excretion of prostaglandin E (Friedman 1978). This may occur consequent to increased production of prostaglandin E2 (Attallah 1979) and from inhibition of prostaglandin E2-9-ketoreductase and prostaglandin E2-15-hydroxy-dehydrogenase in the kidney (Stone 1976; Lin 1978). This effect of furosemide on PG metabolism was the basis for Yeh's randomized trial (Yeh 1982), in which the authors hypothesized that furosemide could affect the severity of nephrotoxicity of indomethacin. Among adults, the diuretic effect of furosemide is blunted by indomethacin in congestive heart failure and cirrhosis (Kahles 1987; Daskalopoulos 1985; Mirouze 1983). In contrast, blunting of furosemide-induced diuresis and natriuresis in healthy volunteers is variable and is not observed in water-loaded subjects (Bergamo 1989).

Rationale and significance of the present review:
Based on investigations by Yeh (1982) and Leititis (1987) summarized in a literature review (Brion 1994, page 830), it has been suggested that indomethacin-induced nephrotoxicity could be decreased by volume expansion for a few hours before its administration or by simultaneous administration of furosemide. Based on the above theoretical considerations, we hypothesized that:

1) Simultaneous administration of furosemide and indomethacin could benefit the premature infant with symptomatic patent ductus arteriosus through mutual antagonism of each drug's side effects (indomethacin-related nephrotoxicity resulting in part from a decrease in renal prostaglandins, vs. furosemide-related increase in prostaglandin E2 possibly leading to a higher incidence of patent ductus arteriosus).

2) Furosemide would decrease indomethacin-related nephrotoxicity in those infants without extracellular volume contraction; conversely, exacerbation of hypovolemia by furosemide could enhance indomethacin nephrotoxicity.

If furosemide cannot prevent indomethacin-related toxicity, or if furosemide increases the risk of failure to close the ductus, slow infusion of indomethacin, extracellular volume expansion or simultaneous administration of dopamine might be more attractive alternatives for clinical practice. In addition, use of other types of prostaglandin synthetase inhibitors such as ibuprofen might be a more attractive alternative for future research.

This review updates the existing review of 'Furosemide in indomethacin-treated infants with symptomatic patent ductus arteriosus' which was published in Cochrane Library (Brion 1998a). The first version of this review was published in abstract form (Brion 1998b) and is in press as a manuscript (Brion 1999).

Objectives

To assess the effects of furosemide in preterm infants treated with indomethacin for a symptomatic patent ductus arteriosus.

The primary objectives were to assess the following:
1) The short-term risk-benefit ratio of furosemide, i.e. whether furosemide affects the incidence of failure of ductal closure and that of indomethacin-related toxicity
2) The effect of furosemide on mid-term and long-term outcome.

The secondary objective was to determine whether the effect of furosemide on renal function and water balance depends on prior extracellular volume.

Criteria for considering studies for this review

Types of studies

Only randomized controlled clinical trials were considered for this review.

Types of participants

Preterm infants with a symptomatic patent ductus arteriosus who were to receive at least one dose of indomethacin.

Types of interventions

Patients were randomly allocated to receive either indomethacin alone or indomethacin preceded by or immediately followed with furosemide.

Types of outcome measures

The study had to include some measure of at least one of the following:

1) Primary outcomes:
a) The short-term risk-benefit ratio of furosemide, i.e. whether furosemide affects the incidence of failure of ductal closure, that of indomethacin-related toxicity including oliguric (< 1 ml/kg/hr) renal failure, symptomatic hyperkalemia (serum K > 7.5 mM/L) and necrotizing enterocolitis, and that of intraventricular hemorrhage and periventricular leukomalacia.
b) The effect of furosemide on mid-term and long-term outcome, including bronchopulmonary dysplasia, duration of oxygen administration, length of stay, pulmonary morbidity during the first year and developmental outcome.

2) Secondary outcome:
To assess whether the effect of furosemide on renal function and water balance depends on prior extracellular volume. The effect of furosemide on the changes in renal function and water balance during indomethacin treatment was compared in three subsets based on initial BUN/creatinine. Variables included the following:
- Urine output
- Blood urea nitrogen (BUN) or urea
- Plasma or serum creatinine concentration
- BUN/creatinine ration
- Glomerular filtration rate (GFR)
- Plasma or serum sodium concentration
- Fractional excretion of sodium
- Osmolal clearance and free water clearance
- Weight

Search strategy for identification of studies

See: Collaborative Review Group search strategy
The standard search method of the Cochrane Neonatal Review Group was used.

1. Published manuscripts:
The standard search method of the Cochrane Neonatal Review Group was used. Computerized literature searches without language restriction using MEDLINE (1966 - 1998), EMBASE (1974 - 1998), Reference Update (1998) in January 1998 were initially conducted. To minimize the risk of missing references, broad-range keywords were used: (Indomethacin and Furosemide) limited to (Clinical trials or Randomized clinical trials). A repeat MEDLINE search was done April 9, 2001 using the keywords (Indomethacin and Furosemide) limited to (Infant, newborn). The Cochrane Controlled Trials Register (CCRT) (The Cochrane Library, Issue 1, 2001) was also searched.

The original MEDLINE search was updated on September 29, 2003 and on April 1 2007, as well as a search of the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 2, 2003 and Issue 1, 2007).

2. Published abstracts:
The abstract books from the following societies were searched:
# American Academy of Pediatrics 1990 - 98 (published in American Journal of Perinatology [1990 - 95] and Pediatrics [1996 - 99]). The sections on Computer and other Technologies, Critical Care and Perinatal Pediatrics were hand searched.
# European Society of Pediatric Research 1991 - 98 (published in Pediatric Research). For years 1991 - 93 and 1997 - 98 the sections on Clinical Trials, Epidemiology, Neonatology, and Pulmonology were hand searched. For 1994 - 96 all abstracts were hand searched.
# Society for Pediatric Research 1985 - 98 (published in Pediatric Research). For years 1985, 88 - 91 and 94 - 98 the following keywords were used: indomethacin, furosemide, and patent ductus arteriosus. For years 1986 - 87 we used the keyword 'anti-inflammatory agents' instead of 'indomethacin'. For years 1992 - 93 (volumes 31 and 33 of Pediatric Research, which do not provide keywords) the sections on Developmental Pharmacology, Neonatology-General, Neonatal Cardiology, Neonatal Pulmonology and Nephrology were hand searched.

3. Database of the Neonatal CRG of the Cochrane Collaboration:
All publications coded under diuretics as intervention in September, 1998 were screened.

4. Selection process:
All studies that involved the administration of indomethacin and furosemide to premature infants with patent ductus arteriosus were considered eligible. Among these eligible studies, the trials that fulfilled the selection criteria cited in the previous section, i.e. randomized allocation to furosemide administration and documented information on one of the described outcome variables, were selected for inclusion in the review.

Methods of the review

The standard method for the Cochrane Collaboration which is described in the Cochrane Collaboration Handbook were used.

Definition of the subsets:
To analyze whether the effect of furosemide depended on extracellular volume, patients were classified into three subsets based on their individual BUN/creatinine ratio before initiating indomethacin. The cutoff values (> 30 vs. 20 - 30 vs. < 20 mg/mg, i.e., > 120 vs. 80 - 120 vs. < 80 mM/mM) were selected a priori based on standard values of pre- vs. intrinsic renal failure (Brion 1994, page 820, table 42 - 8). Attempts to obtain additional information directly from the authors were unsuccessful. Therefore, the mean initial value of BUN/creatinine ratio was used to classify groups of patients (instead of individual patients) into respective pre-determined subsets based on the working hypothesis.

Data collection:
The criteria for entry and exclusion, including the method used to diagnose a hemodynamically significant patent ductus arteriosus (clinical and/or echocardiographic) as well as the methods used for randomization, blinding, and assessment were recorded. Data from the original manuscripts were extracted. Units were homogenized in a standard fashion and standard errors were transformed into standard deviations at two separate occasions; any divergence was resolved by a third independent process. All calculations were done using an electronic spreadsheet.

For each study, comparability of the randomized groups for gestational age, postnatal age, birth weight, present weight, type of environment, fluid intake, urine output and blood chemistry before the first dose of indomethacin was assessed. Information about the dose, rapidity of administration, and route of drug administration was collected.

Outcome variables included the following:
a) Categorical variables:
-Incidence of failure of clinical or echocardiographic closure of the ductus arteriosus in response to one to three doses of indomethacin. For studies including only one dose of indomethacin, patients responding to a second or a third dose were considered as responding to medical therapy, whereas those requiring surgical ligation were considered as failing to respond to medical therapy.
-Incidence of other complications (see above, Section on outcome measures)

b) Continuous variables:
-Change in urine output (ml/kg/hour), plasma creatinine concentration (mg/dl), BUN (mg/dl), BUN/creatinine ratio (mg/mg), glomerular filtration rate (GFR) (ml/min/1.73 m2) assessed by creatinine clearance, fractional excretion of sodium (%), osmolar and free water clearance (ml/min/100 ml of GFR), and serum sodium and potassium (mM/L).
-Duration of oxygen administration
-Number of hospitalizations during the first year
-Developmental outcome

Conversion of values into SI units:
To obtain urea in mM/L, multiply BUN (mg/dl) by 0.36 (or divide by 2.8)
To obtain serum creatinine concentration in micromoles/L, multiply creatinine (mg/dl) by 88
To obtain urea/creatinine in mM/mM, multiply BUN/creatinine (mg/mg) by four

Reviewer agreement:
Selection criteria, search strategy, definition of subgroups and analysis of outcome were agreed upon before initiating the study. Assessment of trials for inclusion and of methodology quality was done independently by each reviewer. There was 100% agreement between the two reviewers.

Statistical methods:
Variables were transformed into the same units and standard errors were transformed into standard deviations, where necessary. For Vargas-Origel's study, Figure 2 was assumed to represent standard errors for all variables, because the error bar for creatinine was identical to the standard error in Table 2. For BUN/creatinine ratio, the square of the coefficient of variation was estimated as the sum of the squares of the coefficients of variation for BUN and for creatinine (Baird 1995). This formula overestimates variability when the values of the numerator and the denominator are related (Armitage 1994). Birth weights, gestational ages, and postnatal ages among pre-determined subsets were compared using one-way analysis of variance (n = 3) or unpaired Student t-test (n = 2), as appropriate. Power analysis was done using the SamplePower program (SPSS, Inc, Chicago, IL).

Changes in continuous variables from baseline to 12 hours after indomethacin administration were compared in the treatment group and the control group. For the percent change in weight after three doses, the standard deviation was provided in the original manuscript. For the other variables, we estimated the variance (var) of change using Follmann's (Follmann 1992) method, described in version 3.0.2 of the Cochrane Collaboration Handbook (page 213):
Var(change)=Var(pretest) + Var(posttest) - 2 x Std(pretest) x Std (posttest) x correlation (pretest, posttest).
A pretest-posttest correlation coefficient of 0.4 was assumed. Sensitivity analysis was conducted by successively using a correlation coefficient of 0.3, 0.4 and 0.5.
Meta-analysis statistics were calculated with MetaView, using a fixed effects approach. Relative risk (RR), risk difference (RD) and weighted mean differences (WMD) are reported with 95% confidence intervals (CI). Of three studies, only one (Romagnoli) involved a full, three-dose course of indomethacin. Data on ductal closure were only available after the end of treatment, whereas data on fluids and renal function after the first dose of indomethacin were available for all studies. Therefore, the meta-analysis of ductal closure was done after the end of treatment and the analysis on renal function after the first dose of indomethacin. The effect of lasix on outcome variables after a full three-dose course of indomethacin (Romagnoli) was also assessed.

Description of studies

The search yielded a total of 10 eligible studies. Seven studies were eliminated because they did not include randomized allocation to furosemide administration. Only 3 studies met eligibility, i.e., they were randomized trials designed to assess the effects of furosemide on defined outcome variables in premature infants with symptomatic patent ductus arteriosus. The total number of patients used in the analysis was 70 (equally distributed between the furosemide group and the control group).

All details of the studies are described in the Table of Included Studies. There were several differences in methods, participants, and interventions among the three studies. One study (Romagnoli) included a three-dose course of indomethacin (interrupted if urine output < 0.5 ml/kg/hr or if serum creatinine > 2 mg/dl), whereas the two other studies (Yeh and Vargas-Origel) included a single dose of indomethacin (with additional doses if required).

In Romagnoli's study the average BUN/creatinine ratio before treatment was more than twice that in the other two studies (~ 25 mg/mg, i.e., 100 mM/mM vs ~ 10 mg/mg, i.e., 40 mM/mM, respectively). Thus, studies were classified into two of the pre-determined three subsets (see Results section).

Post-hoc comparisons: The studies were published over a large time period: Yeh's in 1982, Vargas-Origel's in 1986 and Romagnoli's in 1997, suggesting that there were large differences in treatment between the patient populations and treatments. In Vargas-Origel's study, the average birthweight was 1700 g, compared to 1200 g in the other two studies. The average postnatal age at entry in the furosemide group varied from three days (Romagnoli) to 6.5 days (Vargas-Origel) and 9.5 days (Yeh). Yeh used intravenous administration (duration not provided) and Romagnoli used a 20-minute infusion. In contrast, Vargas-Origel administered furosemide intravenously but indomethacin orally. These authors failed to provide serum concentrations despite the known variability in indomethacin absorption (10 - 90%) in preterm infants (Alpert 1979; Evans 1981; Mrongovius 1982). This might explain in part the large interindividual variability in Vargas-Origel's study.

Methodological quality of included studies

The methodological quality of the studies was evaluated by assessing the risk for four types of bias (selection, performance, attrition and detection) and the method used for assessing ductal patency.

Romagnoli 1997:
Assessment of ductus arteriosus: echocardiographic
Blinding of randomization: can't tell
Blinding of intervention: no
Complete followup: no
Blinding of outcome measurement: can't tell

Vargas-Origel 1986:
Assessment of ductus arteriosus: clinical
Blinding of randomization: can't tell
Blinding of intervention: no
Complete followup: yes
Blinding of outcome measurement: can't tell

Yeh 1982:
Assessment of ductus arteriosus: echocardiographic
Blinding of randomization: yes
Blinding of intervention: no
Complete followup: no
Blinding of outcome measurement: can't tell

In summary, Vargas-Origel's (1986) study had no echocardiographic assessment of the patent ductus arteriosus. All three studies lacked power and blinding. Only Yeh's study provided documentation of allocation concealment. In Romagnoli's study, two patients in the furosemide group developed indomethacin-related toxicity and did not receive the full three-dose indomethacin course. The authors dropped these two patients from the analysis, thereby most likely biasing the results by improving mean renal function after the first dose of indomethacin in the furosemide group, and possibly by reducing the incidence of failure to close the ductus. In contrast, the patient dropped from Yeh's study had an incomplete urine collection; resulting bias is possible but not certain.

Results

Characteristics of the patients included in the meta-analysis:
At study entry there was no significant overall difference between the furosemide-treated infants and controls in birth weight, gestational age, postnatal age, BUN, creatinine, BUN/creatinine ratio, creatinine clearance, fractional excretion of sodium, osmolar clearance, free water clearance, and serum sodium and potassium concentrations. However, urine output before therapy was lower in the furosemide group than in controls, WMD -0.65 ml/kg/hour (95% CI -1.29, -0.004). In addition, in Romagnoli's study, patients randomized to furosemide had a lower birth weight, gestational age and baseline urine output than those in the control group.

Studies were classified into two subsets: Subset one, average BUN/creatinine ratio 20 - 30 mg/mg (i.e., 80 - 120 mM/mM) included Romagnoli's study, while subset two, average BUN/creatinine ratio < 20 mg/mg (i.e., < 80 mM/mM), included Vargas-Origel's and Yeh's studies. Initial BUN/creatinine ratio in subset one was twice that in subset 2 (25.6 ± 10.8 vs. 10.3±5.6 mg/mg, p < 0.001).

Post-hoc analyses: Patients in subset one had a significantly lower gestational age (28.4 ± 1.9 weeks, n = 34, vs. 31.6 ± 2.9 weeks, n = 36, p < 0.001), birth weight (1126 ± 269 g vs. 1412 ± 448 g, p < 0.01) and postnatal age at initiation of therapy (3.5 ± 2.7 days vs. 7.7 ± 6.6 days, p < 0.01) than those in subset two. Studies in subset one were published a decade later than those in subset two.

FUROSEMIDE VS. CONTROL (PROTOCOL INCLUDING 1-3 DOSES) (Comparison 01)
Effects of indomethacin on the incidence of failure of ductal closure:

In Romagnoli's study, 7 of 16 patients in the furosemide group and 6 of 18 patients in the control group failed to respond to indomethacin. Two of 18 patients randomized to furosemide did not receive a complete course and were excluded from the analysis. In Vargas-Origel's study, no patient in either group initially failed to respond to indomethacin. However the ductus reopened in two patients in the furosemide group and one in the control group. Two patients required and responded to a second dose of indomethacin and the third patient, assigned to the furosemide group, required surgical ligation. Thus, one of nine patients failed to respond to one to two doses of indomethacin in the furosemide group and none of eight in the control group. In Yeh's study, 3 of 10 patients in the furosemide group and 2 of 9 in the control group failed to respond to one dose of indomethacin; in the control group the ductus reopened in one patient. Thus, 3/10 patients in the furosemide group and 3/9 in the control group failed to respond to one dose of indomethacin.

Furosemide administration did not significantly increase the risk of failure of ductal closure after indomethacin therapy [pooled relative risk 1.25 (CI 0.62, 2.52) and risk difference 0.07 (CI -0.14, +0.28)]. A sensitivity (intent-to-treat basis) analysis, assuming that the two patients excluded in the analysis in Romagnoli's study had failed to respond to a partial indomethacin course, yielded a relative risk of 1.37 (CI 0.70, 2.67) and a risk difference of 0.10 (CI -0.11, +0.31). In view of the small total number (n = 70) of enrolled patients, it is therefore possible that furosemide administration might increase the risk of failure of ductal closure after a single dose of indomethacin by as much as 31% (absolute risk difference) or decrease it by 14%.

Effects of furosemide on indomethacin toxicity and other complications:
Two patients in Romagnoli's study developed indomethacin toxicity and were removed from the analysis, presumably because of oliguria or high plasma creatinine concentration. In the same study, approximately half the patients in each group were found to have an intraventricular hemorrhage > 2, according to Papile classification.

In Yeh's study, one patient with oliguria at study entry had an increase in urine output and GFR after furosemide and indomethacin. Within 12 - 24 hours after a single dose of indomethacin, patients on furosemide had twice as much urine output as controls (2.1 ± 0.9 ml/kg/hr vs. 1.0 ± 0.9 ml/kg/hr, p < 0.05). Because of the low mean urine output in controls, at least some must have been oliguric; however, the number is not provided. At that time average creatinine clearance was identical in the two groups, suggesting that oliguria was unlikely to correspond to renal failure. No information is provided about the incidence of other important short-, mid- or long-term outcome variables.

Effects of indomethacin on renal function and fluid and electrolytes:
Furosemide administration significantly increased urine output (WMD 1.7 ml/kg/hour, CI 0.9, 2.5) regardless of initial BUN/creatinine ratio. Sensitivity analysis showed that CI was minimally affected by increasing the value of pretest-posttest correlation coefficient from 0.3 to 0.5. For instance, urine output WMD CI was 0.8, 2.5; 0.9, 2.5 and 1.0, 2.4 ml/kg/hr for a correlation coefficient of 0.3, 0.4 and 0.5, respectively.

The effect of furosemide on creatinine clearance was not statistically heterogeneous among studies (chi-square value 3.82, df = 2). Nevertheless, in subset two, a single dose of furosemide significantly increased creatinine clearance (WMD 3.3 ml/min/1.73 m2, CI 0.7, 6.0), whereas it did not affect creatinine clearance in subset one. These differences were not affected by changing the pretest-posttest correlation from 0.4 to 0.3.

Furosemide increased fractional excretion of sodium (WMD 3.4 %, CI 1.8, 4.9); the effect was observed in all studies. The effect of furosemide on osmolal clearance was significantly different among studies, as shown by statistical heterogeneity (chi-square value 7.90, p < 0.05, df = 2).

Furosemide significantly increased osmolal clearance in Yeh's study but not in the other studies. Because individual data are unavailable, we were unable to analyze possible reasons for these differences.

FUROSEMIDE VS. CONTROL (PROTOCOL INCLUDING 3 DOSES (Comparison 02):

In Romagnoli's study, the average BUN/creatinine ratio before randomization was > 20 mg/mg (i.e., > 80 mM/mM). Furosemide administration did not significantly affect the risk for failure of ductal closure. However, as mentioned above, two patients in the furosemide group were excluded from the analysis because of indomethacin toxicity. The treatment effect of three doses of furosemide included the following:
(1) A greater increase in urine output (+0.1 ± 1.2 ml/kg/hour vs. -1.4 ± 1.5 ml/kg/hour, p < 0.01, WMD 1.5, CI 0.6, 2.4)
(2) A greater increase in BUN (11.0±12.6 mg/dl vs 1.0±12.9 mg/dl, p<0.05, WMD 10.0, CI 1.4,18.6); trends towards higher serum creatinine concentration and BUN/creatinine did not reach statistical significance.
(3) A greater increase in fractional excretion of sodium (WMD 2.7%, CI 0.4, 5.0) and osmolal clearance (WMD 0.24 ml/min/100 ml GFR, CI 0.08, 0.40)
(4) An increase in percent weight loss, shown in MetaView as a more negative percentage weight change from baseline (-5.35 ± 6.77% vs -0.68 ± 5.70%, p<0.05, WMD -4.67, CI -8.91, -0.44).

Discussion

1. Effect of furosemide on the incidence of failure of ductal closure:
Although the tendency toward a slight (7%) increase in risk for failure of ductal closure did not reach statistical significance, the number of patients enrolled was clearly insufficient to rule out a 0.1 risk reduction or a 0.3 risk increase. If this were a single randomized trial including 70 patients, the power would have been only 20% for detecting a 0.1 risk reduction, 15% for detecting a 0.1 risk increase and 75% for detecting a 0.3 risk increase (assuming similar rate of failure of ductal closure in the control group, two-tailed chi-square analysis and an alpha error of 0.05). Thus, a much larger randomized trial would be needed to rule out a significant increase in failure of ductal closure.

2. Effect of furosemide on renal function and water balance:
All studies in this review showed that furosemide increased urine output compared to controls, regardless of the initial BUN/creatinine ratio. In Romagnoli's study, the furosemide-induced increase in urine output persisted throughout the three-dose indomethacin course, and was associated with a weight loss not observed in the control group.

Furosemide improved glomerular function in patients with an initial BUN/creatinine ratio < 20 mg/mg (< 80 mM/mM), but not in those with an initial BUN/creatinine ratio 20 - 30 mg/mg (80 - 120 mM/mM). In the latter patients, serum creatinine concentration increased from 1.2 to 1.4 mg/dl after 3 doses of indomethacin. It is possible that the two drop-outs in the furosemide group may have developed renal failure or oligoanuria after indomethacin, thereby minimizing the difference between the two subsets. These data suggest that a single dose of furosemide may improve glomerular filtration rate in indomethacin-treated patients with normal hydration at study entry (presumably by mutual antagonism of side effects of furosemide and indomethacin) but not in those with dehydration (presumably because of worsening hypovolemia and indomethacin nephrotoxicity).

3. Comments about the studies available for this review:
Original studies available for this review had at least possible if not likely bias. None of the studies available for this review had a perfect design. Two of them lacked allocation concealment, no information was available about blinding of drug administration and analysis of the data, and two had incomplete followup. Thus, four types of bias are possible: selection, performance bias, attrition and detection. Attrition bias is very likely in Romagnoli's study. Furthermore, Vargas' study lacked echocardiographic confirmation of the patent ductus arteriosus and did not use parenteral administration of indomethacin. This could have resulted in errors in diagnosis of patent ductus arteriosus, in assessment of therapeutic response and in erratic absorption of indomethacin.

Insufficient or no data were available to assess the number of patients who developed indomethacin-related complications such as oliguria or necrotizing enterocolitis, or other important outcomes such as intraventricular hemorrhage, periventricular leukomalacia, bronchopulmonary dysplasia, duration of ventilation, or long-term outcome.

4. Comments about this systematic review:
This review was conducted using the guidelines of the Cochrane Collaboration. Nevertheless, its significance is limited by the weakness and heterogeneity of available studies, by the small total number of patients, by lack of data on individual patients, and by changes in therapy after these studies have been performed.

Heterogeneity included study design, entry criteria, quality of assessment of ductal patency, protocol of administration, route of administration, and number of doses of indomethacin. Finally, the studies spanned more than a decade (1982-1997), during which time several changes occurred in neonatal practice including ventilation methods, prenatal steroids, surfactant and timing of indomethacin administration. In contrast with the other two studies, Vargas' study included only clinical assessment of ductal patency, oral administration of indomethacin, and patients with a higher gestational age and a higher birth weight. Therefore, the meta-analysis was repeated after removing this study. Results of this sensitivity analysis were virtually identical to the original one, except for a significant increase in osmolal clearance in the second subset: WMD 3.3 (CI 0.7,5.9). Otherwise, this post-hoc analysis yielded exactly the same significant differences as the analysis using all 3 studies.

Meta-analysis of failure of ductal closure was done after the end of treatment, ranging from one to three doses of indomethacin. Nevertheless, results were similar in studies assessing the response to one to two doses and in assessing the response after three doses of indomethacin. Meta-analysis of renal function was done after the first dose of indomethacin only. Administration of a single dose of indomethacin is at variance with current practice (see above paragraph). Nevertheless, in Romagnoli's study, results after one dose of indomethacin showed trends and differences that were similar to those after the full course.

Individual information on BUN/creatinine ratio before the first dose of indomethacin was not available. For this reason, subset analysis was done by comparing studies with different average BUN/creatinine ratios, rather than by comparing subsets of patients in each study who had BUN/creatinine ratios within the respective predetermined ranges. Post-hoc analyses revealed additional differences (e.g., year of publication, gestational age, postnatal age) between subsets; therefore, it is possible that some of these statistically significant post-hoc comparisons could explain in part differences in outcome. In addition, a high BUN/creatinine ratio (in Romagnoli's study) may have resulted not only from low extracellular volume but also from catabolism, from high amino acid load (high intake or gastrointestinal hemorrhage), or from low muscle mass (causing a low serum creatinine concentration). For this review, subsets based on other estimates of extracellular volume, e.g. weight loss from birth until initiation of indomethacin, could not be performed because this variable was available in only one study.

All these factors limit the strength of inferences that can be made from this systematic review. Additional, well designed studies with adequate power and blinding would be required to confirm or refute the present findings.

The increase in urine output and in glomerular filtration rate in subset two observed in the present review cannot be taken as enough evidence to support use of furosemide in well-hydrated preterm infants receiving indomethacin for symptomatic patent ductus arteriosus. The implications of the results are limited by the availability of other methods to prevent indomethacin toxicity without risk to increase the failure to close the ductus arteriosus. Normal renal and cerebral blood flow can most often be maintained by using slow rate of infusion of indomethacin (Christmann 1998; Zecca 1994), and renal complications of indomethacin can be limited by extracellular volume expansion (Leititis 1987) or dopamine infusion (Seri 1984).

Reviewers' conclusions

Implications for practice

Original studies available for this review are weak, with at least possible if not likely bias, and with small sample size. No or minimal data were provided about important outcomes such as oliguric renal failure, intraventricular hemorrhage, bronchopulmonary dysplasia, duration of ventilation, necrotizing enterocolitis or long-term outcome. Two of three studies were performed more than 10 years ago, suggesting that care practices may have changed dramatically.

This systematic review does not provide enough conclusive evidence to support the administration of furosemide to premature infants treated with indomethacin for symptomatic patent ductus arteriosus.

In view of the small total number of enrolled patients, it is possible that furosemide administration might increase the risk of failure of ductal closure after a single dose of indomethacin by as much as 31% (absolute risk difference) or decrease it by 14%.
Furosemide increased urine output in all three studies, regardless of the initial BUN/creatinine ratio, and increased creatinine clearance only in patients with initial BUN/creatinine ratio < 20 mg/mg. Thus, dehydration appears to be a contraindication for furosemide administration in premature infants treated with indomethacin for symptomatic patent ductus arteriosus.

Implications for research

Whether furosemide is a useful additive to indomethacin therapy in well-hydrated preterm infants with symptomatic patent ductus arteriosus could only be assessed by a multicenter randomized clinical trial with adequate blinding. The study should include at least 413 patients in each group to have of 90% power for detecting a 0.1 risk increase for failure to close the ductus (using two-tailed chi-square analysis, assuming a 22% failure to close the ductus in controls, and an alpha error of 0.05). The study should use the current method of indomethacin administration, i.e., three doses given by slow or continuous infusion. Such a study may not be indicated if current studies using slow or continuous indomethacin infusion or ibuprofen show good efficacy and limited toxicity in large numbers of patients.

Acknowledgements

Potential conflict of interest

None

Characteristics of included studies

Study	Methods	Participants	Interventions	Outcomes	Notes	Allocation concealment
Romagnoli 1997	Blinding of randomization: can't tell; blinding of intervention: no; complete followup: no; blinding of outcome measurement: can't tell Parallel design without cross-overs. Two patients in group 1 did not complete the study because of indomethacin side effects and were excluded from the analysis by the authors. Whether these side effects were renal, gatrointestinal or cerebral is not mentioned in the manuscript. Fluid intake in both groups followed an indentical protocol (60 ml/kg on first day, 70 on the second day, 80 on the third, 100 on the fourth and up to 150 at the end of the first week of life).	A total of 36 preterm infants with birthweight <1750 grams with hemodynamically significant patent ductus arteriosus confirmed by color Doppler echocardiography were entered into the study. The final numbers include 16 patients on furosemide and 18 in the control group. The average birth weight was 1088�0 grams in group 1 and 1159�8 grams in the control group; the average gestational age was 27.9�0 weeks and 28.9�9 weeks, respectively. All patients had respiratory distress syndrome, and were on the ventilator. All patients in group 1 and 72% in group 2 received surfactant. The average BUN/creatinine ratio before therapy was 26.7�.2 mg/mg in group 1 and 24.5�.3 mg/mg in group 2. Treatment was started at 3.0�7 days in group 1 and 3.9�4 days in controls.	Patients in both groups received three doses of indomethacin 0.2 mg/kg i.v. every 12 hours. Patients in group 1 also received furosemide 1 mg/kg immediately after each dose of indomethacin. Indomethacin was infused over a 20-minute period. The treatment was suspended if urine output was < 0.5 ml/kg/hour or if plasma creatinine concentration was > 2 mg/dl.	There was no difference between the two groups in therapeutic effectiveness (i.e., closure of the ductus arteriosus confirmed by echocardiography, 9/16 vs. 12/18 in groups I and 2, respectively) or in weight, serum electrolytes and osmolarity. During treatment, patients in group 1 but not those in group 2 developed a significant increase in blood urea nitrogen and plasma creatinine (which peaked at 12 hours after the third dose) and in glomerular filtration rate (which had a trough 12 hours after the first dose). Patients in group 1 had a higher fractional excretion of sodium and potassium and a higher osmolal clearance and free water clearance than those in group 2. Patients in group 2 but not those in group 1 had a decrease in urine output.	In group 1 (furosemide), the maximum decrease in glomerular filtration rate and increase in fractional excretion of sodium were observed 12 hours after the first dose; the maximum increase in plasma creatinine developed 12 hours after the third dose.	B
Vargas-Origel 1986	Blinding of randomization: can't tell; blinding of intervention: no; complete followup: yes; blinding of outcome measurement: can't tell. Parallel design without cross-overs; no information recorded about blinding. No information is available about number of patients excluded from the study. Fluid intake was 100 ml/kg/day in both groups, and sodium intake was 2 mEq/kg/day.	A total of 17 premature infants with gestational age less than 36 weeks with hemodynamically significant patent ductus arteriosus (not confirmed by echocardiography) were entered into the study. Patients were excluded if they had thrombocytopenia (<75,000/mm3), severe hyperbilirubinemia possibly requiring exchange transfusion, necrotizing enterocolitis, gastrointestinal bleeding, renal failure (plasma creatinine concentration > 1.5 mg/dl). The final number of patients was 9 in the furosemide group and 8 in the control group. The average gestational age was 32.6�4 weeks in the furosemide group and 32.8�7 weeks in controls; respective weights were 1650�0 and 1751�0 grams. Four of 9 patients were on the ventilator in the furosemide group, compared to 6 of 8 in the control group. Postnatal age at the time of entry into the study was 6.5�6 and 3.4�8 days, respectively. BUN/creatinine ratio before therapy was 9.3�7 mg/mg in group 1 and 9.9�7 mg/mg in group 2.	Patients in both groups received one oral dose of indomethacin 0.2 mg/kg and only if necessary received a second dose. Patients in group 1 but not in group 2 received one i.v. dose of furosemide 1 mg/kg.	One patient in each group required a second dose of indomethacin; one patient in group 1 required surgical ligation for reopening of the ductus. There was no difference in serum creatinine and electrolytes, fractional excretion of sodium and osmolar clearane, water clearance and creatinine clearane. Patients in group 1 had a higher urine output than those on placebo during the first 12 hours after indomethacin.	Only two patients needed more than one dose of indomethacin. Maximum changes in blood urea nitrogen were observed in both groups at 12 hours after indomethacin. In group 1, maximum changes in urine output and in osmolar and free water clearances were observed 12 hours after treatment, and maximum change in fractional excretion of chloride was observed after 12-24 hours after treatment.	B
Yeh 1982	Blinding of randomization: yes; blinding of intervention: no; complete followup: no; blinding of outcome measurement: can't tell Parallel design, no cross-over. An independent observer kept twenty envelopes which were used for randomization. One patient was dropped from the study because of incomplete urine collection. Fluid intake was similar in both groups, 120 ml/kg/day (150 if phototherapy).	Twenty premature infants with a significant patent ductus arteriosus (diagnosed clinically and confirmed by echocardiogram) were randomized. The final number of patients was 10 in the furosemide group and 9 in the control group. The average birth weight was 1.19�32 kg and 1.12�39 kg in the furosemide group and the placebo group, respectively; gestational ages were, respectively, 30.7�5 and 30.4�7 weeks. Nine patients in each group required positive pressure support. Patients were entered into the study at a mean postnatal age of 9.5�4 and 10.7�.2 days. Average BUN/creatinine ratio before indomethacin was 11.7�1 mg/mg in group 1 and 10.9�0 mg/mg in group 2.	Patients in both groups received one single indomethacin dose of 0.3 mg/kg iv, whereas those in group 1 only received in addition furosemide 1 mg/kg i.v. immediately after indomethacin. There is no information about the rapidity of indomethacin infusion.	Seven infants in each group initially responded to indomethacin (closure of the ductus) but one had recurrence in group 2. Patients in group 1 had higher urine output, glomerular filtration rate, fractional excretion of sodium and of chloride, and osmolar clearance and free water clearance than those in the placebo group. One infant in the furosemide group who had oliguria of presumably pre-renal origin responded to furosemide and indomethacin with normalization of urine output, doubling of glomerular filtration rate and increase in fractional excretion of sodium.	In this review, groups 1 and 2 were switched compared to the original publication, so that group 1 corresponds to the furosemide group, as in the other studies included in the review. Maximum changes in blood urea nitrogen, glomerular filtration rate and fractional excretions of sodium and chloride from baseline was observed 12 hours after indomethacin. Urine output remained low in the control group for up to 24 hours after treatment.	A

Characteristics of excluded studies

Study	Reason for exclusion
Barrington 1994	No randomized allocation to furosemide administration
Maslarska 1997	No randomized allocation to furosemide administration
Merritt 1981	No randomized allocation to furosemide administration
Neubauer 1995	No randomized allocation to furosemide administration
Petion 1990	No randomized allocation to furosemide administration
Ransom 1990	No randomized allocation to furosemide administration
Yeh 1986	No randomized allocation to furosemide administration

References to studies

References to included studies

Romagnoli 1997 {published data only}

Romagnoli C, Zecca E, Papacci P, De Carolis MP, Giannini R, Gallini F, et. al. Furosemide does not prevent indomethacin-induced renal side effects in preterm infants. Clinical Pharmacology and Therapeutics 1997;62:181-6.

Vargas-Origel 1986 {published data only}

Vargas-Origel A, Cruz-Anguiano V, L髉ez-Monta駉 A. Indometacina y furosemide en el cierre del conducto arterioso. Boletin Medico del Hospital Infantil de Mexico 1986;43:482-8.

Yeh 1982 {published data only}

Yeh TF, Wilks A, Singh J, Betkerur M, Lilien L, Pildes RS. Furosemide prevents the renal side effects of indomethacin therapy in premature infants with patent ductus arteriosus. Journal of Pediatrics 1982;101:433-7.

References to excluded studies

Barrington 1994 {published data only}

Barrington KJ, Fox M. Predicting oliguria following indomethacin for treatment of patent ductus arteriosus. American Journal of Perinatology 1994;11:220-2.

Barrington KJ. Predicting oliguria following indomethacin. Pediatric Research 1992;31:194A.

Maslarska 1997 {published data only}

Maslarska R, Spasov L, Klyasova R, Tomova S. Therapeutic conduct in preterm neonates with patent ductus arteriosus persistens and respiratory distress syndrome (RDS). Pediatriya 1997;36:22-3.

Merritt 1981 {published data only}

Merritt TA, Harris JP, Roghmann K, Wood B, Campanella V, Alexson C, et. al. Early closure of the patent ductus arteriosus in very-low-birth-weight infants: A controlled trial. Journal of Pediatrics 1981;99:281-6.

Neubauer 1995 {published data only}

Neubauer AP, M黮ler-Deile K. Verschluss des h鋗odynamisch wirksamen persistierenden Ductus Botalli (PDA) beim Fr黨geborenen unter 1000 G mit Indometacin. Monatsschrift fur Kinderheilkunde 1995;143:1224-30.

Petion 1990 {published data only}

P閠ion AM, Gouyon JP, Sandr� D. Le furos閙ide pr関ient-il les perturbations des fonctions r閚ales induites par l'administration d'indom閠acine chez le pr閙atur� Archives Francaises de P閐iatrie 1990;47:615-6.

Ransom 1990 {published data only}

Ransom JL, Gal P, Weaver RL, Schall S, Brown Y. Impact of concurrent furosemide on indomethacin pharmacokinetics Re: Efficacy for patent ductus arteriosus closure and renal toxicity. American Journal of Perinatology 1990;7:393.

Yeh 1986 {published data only}

Yeh TF, Wilks A, Luken J, Pildes RS. Indomethacin therapy in premature infants with patent ductus arteriosus and oliguria. Developmental Pharmacology and Therapeutics 1986;9:369-74.

Yeh TF, Wilks A, Raval D, Pildes RS. Furosemide prevents the renal side effects of indomethacin in premature infants with PDA and oliguria. Pediatric Research 1985;19:181A.

* indicates the primary reference for the study

Other references

Additional references

Alpert 1979

Alpert BS, Lewins MJ, Rowland DW, Grant MJ, Olle PM, Soldin SJ, et. al. Plasma indomethacin levels in preterm newborn infants with symptomatic patent ductus arteriosus - clinical and echocardiographic assessments of response. Journal of Pediatrics 1979;95:578-82.

Armitage 1994

Armitage P, Berry G. Sampling. In: Armitage P, Berry G, editor(s). Statistical Methods in Medical Research. Third edition. Oxford: Blackwell Scientific Publications, 1994:78-92.

Attallah 1979

Attallah A, Stahl R, Bloch D, Lee JB. Furosemide stimulates and saralasin inhibits renal prostaglandin E2 biosynthesis. Clinical Research 1979;27:559A.

Baird 1995

Baird DC. Statistics of observation. In: Baird DC, editor(s). Experimentation. An introduction to measurement theory and experimental design. New Jersey: Prentice Hall, 1994:29-56.

Bergamo 1989

Bergamo RR, Cominellli F, Kopple JD, Zipser RD. Comparative acute effects of aspirin, diflunisal, ibuprofen and indomethacin on renal function in healthy man. American Journal of Nephrology 1989;9:460-3.

Brion 1994

Brion LP, Satlin LM, Edelmann CM Jr. Renal disease. In: Avery GB, Fletcher MA, MacDonald MG, editor(s). Neonatology: Pathophysiology and management of the newborn. Philadelphia: JB Lippincott Co, 1994:792-886.

Christmann 1998

Christmann V, Semmekrot BA, Van de Bor M. Changes in cerebral, renal and mesenteric blood flow velocity during continuous and bolus infusion of indomethacin [abstract]. Pediatric Research 1998;43:169A.

Cochran 1989

Cochran J, Reddy R, Devaskar U. Effect of indomethacin (I) vs I + dopamine on serum BUN and creatinine, urinary output and the closure of patent ductus arteriosus in preterm neonates with hyaline membrane disease. Pediatric Research 1989;25:211A.

Daskalopoulos 1985

Daskalopoulos G, Kronborg I, Katkov W, Gonzalez M, Laffi G, Zipser RD. Sulindac and indomethacin suppress the diuretic action of furosemide in patients with cirrhosis and ascites: evidence that sulindac affects renal prostaglandins. American Journal of Kidney Diseases 1985;6:217-21.

Evans 1981

Evans M, Bhat R, Vidyasagar D, Patel M, Hastreiter A. A comparison of oral and intravenous indomethacin dispositions in the premature infant with patent ductus arteriosus. Pediatric Pharmacology 1981;1:251-8.

Follmann 1992

Zimmerman JJ. Bronchoalveolar inflammatory pathophysiology of bronchopulmonary dysplasia. Clinics in Perinatology 1995;22:429-56.

Fowlie 1997

Fowlie PW. Prophylactic intravenous indomethacin in very low birth infants. In: Cochrane Database of Systematic Reviews, Issue 3, 1998.

Friedman 1982

Friedman CA, Parks BR, Rawson JE, Serwer GA, Anderson PAW. Indomethacin and the preterm infant with a patent ductus arteriosus: Relationship between plasma concentration and ductus closure. Developmental Pharmacology and Therapeutics 1982;4:37-46.

Green 1983

Green TP, Thompson TR, Johnson De, Lock JE. Furosemide promotes patent ductus arteriosus in premature infants with the respiratory-distress syndrome. New England Journal of Medicine 1983;308:743-8.

Kahles 1987

Kahles H, Riegger AJ. Indometacin und Furosemid bei Patienten mit Herzinsuffizienz. Nierenfunktion, Renin-Angiotensin-System und renale Prostaglandine. Deutsche Medizinische Wochenschrift 1987;112:1737-40.

Leititis 1987

Leititis JU, Burghard R, Gordjani N, Wildberg A, Seyberth HW, Brandis M. Effect of a modified fluid therapy on renal function during indomethacin therapy for persistent ductus arteriosus. Acta Paediatrica Scandinavica 1987;76:789-94.

Lin 1978

Lin YM, Jarabak J. Isolation of two proteins with 9-ketoprostaglandin reductase and NADP-linked 15-hydroxyprostaglandin dehydrogenase activities and studies on their inhibition. Biochemical and Biophysical Research Communications 1978;81:1227-34.

Lin 1995

Lin YJ, Tsai YJ, Chen JS, Lin JS, Wu JM, Lin CH, et. al. Renal effects and urinary excretion of prostaglandin following indomethacin therapy in premature infants with patent ductus arteriosus. Acta Paediatrica Sinica 1995;36:104-7.

Mirouze 1983

Mirouze D, Zipser RD, Reynolds TB. Effect of inhibitors of prostaglandin synthesis on induced diuresis in cirrhosis. Hepatology 1983;3:50-5.

Mrongovius 1982

Mrongovius R, Imbeck H, Wille L, Muller H, Seyberth HW. Variability of serum indomethacin concentrations after oral and intravenous administration to preterm infants. European Journal of Pediatrics 1982;138:151-3.

Seri 1984

Seri I, Tulassay T, Kiszel J, Csomor S. The use of dopamine for the prevention of the renal side effects of indomethacin in premature infants with patent ductus arteriosus. International Journal of Pediatric Nephrology 1984;5:209-14.

Seyberth 1983

Seyberth HW, Rascher W, Hackenthal R, Wille L. Effect of prolonged indomethacin therapy on renal function and selected vasoactive hormones in very-low-birth-weight infants with symptomatic patent ductus arteriosus. Journal of Pediatrics 1983;103:979-84.

Stone 1976

Stone KJ, Hart M. Inhibition of renal PGE2-9-ketoreductase by diuretics. Prostaglandins 1976;12:197-207.

van Bel 1991

van Bel F, Guit GL, Schipper J, van de Bor M, Baan J. Indomethacin-induced changes in renal blood flow velocity waveform in premature infants investigated with color Doppler imaging. Journal of Pediatrics 1991;118:621-6.

Zecca 1994

Zecca E, Papacci P, De Carolis MP, Pasquini R, Vento G, Romagnoli C. Il trattamento dell pervieta del dotto arterioso con indometacina nel neonato pretermine: confronto tra due diverse modalita di infusione. Riv Ital Pediatr (IJP) 1994;20:228-33.

Other published versions of this review

Brion 1998a

Brion LP, Campbell DE. Furosemide in indomethacin-treated infants with symptomatic ductus arteriosus. In: Cochrane Database of Systematic Reviews, Issue 3, 1998.

Brion 1998b

Brion LP, Campbell DE. Furosemide in indomethacin-treated premature infants with symptomatic patent ductus arteriosus - Systematic review and meta-analysis. In: Journal of the American Society of Nephrology. Vol. 9. 1998:141A.

Brion 1999

Brion LP, Campbell DE. Furosemide in indomethacin-treated infants. Systematic review and meta-analysis. Pediatric Nephrology 1999;13:212-8.

Brion 2001

Brion LP, Campbell DE. Furosemide for prevention of morbidity in indomethacin-treated infants with patent ductus arteriosus. In: Cochrane Database of Systematic Reviews, Issue 3, 2001.

Brion 2004

Brion LP, Campbell DE. Furosemide for prevention of morbidity in indomethacin-treated infants with patent ductus arteriosus. In: Cochrane Database of Systematic Reviews, Issue 1, 2004.

Comparisons and data

Comparison or outcome	Studies	Participants	Statistical method	Effect size
01 Furosemide vs. control (protocol including 1-3 doses, see text)
01 Failure of closure of the ductus arteriosus	3	70	RR (fixed), 95% CI	1.25 [0.62, 2.52]
02 Change in urine output (ml/kg/hr)	3	70	WMD (fixed), 95% CI	1.69 [0.90, 2.48]
03 Change in BUN (mg/dl)	3	70	WMD (fixed), 95% CI	-0.93 [-5.53, 3.67]
04 Change in serum creatinine (mg/dl)	3	70	WMD (fixed), 95% CI	-0.01 [-0.14, 0.12]
05 Change in BUN/creatinine (mg/mg)	3	70	WMD (fixed), 95% CI	-0.55 [-5.05, 3.95]
06 Change in creatinine clearance (ml/min/1.73 m2)	3	70	WMD (fixed), 95% CI	1.41 [-0.45, 3.26]
07 Change in fractional excretion of sodium (%)	3	70	WMD (fixed), 95% CI	3.39 [1.83, 4.95]
08 Change in osmolal clearance (ml/min/100 ml GFR)	3	70	WMD (fixed), 95% CI	0.27 [0.07, 0.47]
09 Change in free water clearance (ml/min/100 ml GFR)	3	70	WMD (fixed), 95% CI	0.11 [-0.02, 0.25]
10 Change in serum Na (mM/L)	3	70	WMD (fixed), 95% CI	0.60 [-2.03, 3.24]
11 Change in serum K (mM/L)	2	53	WMD (fixed), 95% CI	-0.01 [-0.38, 0.35]
02 Furosemide vs. control (protocol including 3 doses; initial BUN/creatinine 20-30 mg/mg)
01 Failure of closure of the ductus arteriosus	1	34	RR (fixed), 95% CI	1.31 [0.56, 3.09]
02 Complication preventing completion of indomethacin course	1	36	RR (fixed), 95% CI	5.00 [0.26, 97.37]
03 Change in urine output (ml/kg/hr)	1	34	WMD (fixed), 95% CI	1.50 [0.57, 2.43]
04 Change in BUN (mg/dl)	1	34	WMD (fixed), 95% CI	10.00 [1.42, 18.58]
05 Change in serum creatinine (mg/dl)	1	34	WMD (fixed), 95% CI	0.20 [-0.03, 0.43]
06 Change in BUN/creatinine (mg/mg)	1	34	WMD (fixed), 95% CI	3.14 [-5.70, 11.98]
07 Change in creatinine clearance (ml/min/1.73 m2)	1	34	WMD (fixed), 95% CI	-0.40 [-3.87, 3.07]
08 Change in fractional excretion of sodium (%)	1	34	WMD (fixed), 95% CI	2.70 [0.39, 5.01]
09 Change in osmolal clearance (ml/min/100 ml GFR)	1	34	WMD (fixed), 95% CI	0.24 [0.08, 0.40]
10 Change in free water clearance (ml/min/100 ml GFR)	1	34	WMD (fixed), 95% CI	0.02 [-0.11, 0.15]
11 Change in serum Na (mM/L)	1	34	WMD (fixed), 95% CI	0.00 [-3.16, 3.16]
12 Change in serum K (mM/L)	1	34	WMD (fixed), 95% CI	-0.40 [-0.85, 0.05]
13 Weight change (grams)	1	34	WMD (fixed), 95% CI	-54.00 [-237.98, 129.98]
14 Percent weight change (%)	1	34	WMD (fixed), 95% CI	-4.67 [-8.91, -0.43]

01 Furosemide vs. control (protocol including 1-3 doses, see text)

01.01 Failure of closure of the ductus arteriosus

01.01.01 BUN/creatinine 20-30 mg/mg

01.01.02 BUN/creatinine <20 mg/mg

01.02 Change in urine output (ml/kg/hr)

01.02.01 Initial BUN/creatinine 20-30 mg/mg

01.02.02 Initial BUN/creatinine <20 mg/mg

01.03 Change in BUN (mg/dl)

01.03.01 Initial BUN/creatinine 20-30 mg/mg

01.03.02 Initial BUN/creatinine <20 mg/mg

01.04 Change in serum creatinine (mg/dl)

01.04.01 Initial BUN/creatinine 20-30 mg/mg

01.04.02 Initial BUN/creatinine <20 mg/mg

01.05 Change in BUN/creatinine (mg/mg)

01.05.01 Initial BUN/creatinine 20-30 mg/mg

01.05.02 Initial BUN/creatinine <20 mg/mg

01.06 Change in creatinine clearance (ml/min/1.73 m2)

01.06.01 Initial BUN/creatinine 20-30 mg/mg

01.06.02 Initial BUN/creatinine <20 mg/mg

01.07 Change in fractional excretion of sodium (%)

01.07.01 Initial BUN/creatinine 20-30 mg/mg

01.07.02 Initial BUN/creatinine < 20 mg/mg

01.08 Change in osmolal clearance (ml/min/100 ml GFR)

01.08.01 Initial BUN/creatinine 20-30 mg/mg

01.08.02 Initial BUN/creatinine <20 mg/mg

01.09 Change in free water clearance (ml/min/100 ml GFR)

01.09.01 Initial BUN/creatinine 20-30 mg/mg

01.09.02 Initial BUN/creatinine <20 mg/mg

01.10 Change in serum Na (mM/L)

01.10.01 Initial BUN/creatinine 20-30 mg/mg

01.10.02 Initial BUN/creatinine <20 mg/mg

01.11 Change in serum K (mM/L)

01.11.01 Initial BUN/creatinine 20-30 mg/mg

01.11.02 Initial BUN/creatinine <20 mg/mg

02 Furosemide vs. control (protocol including 3 doses; initial BUN/creatinine 20-30 mg/mg)

02.01 Failure of closure of the ductus arteriosus

02.02 Complication preventing completion of indomethacin course

02.03 Change in urine output (ml/kg/hr)

02.04 Change in BUN (mg/dl)

02.05 Change in serum creatinine (mg/dl)

02.06 Change in BUN/creatinine (mg/mg)

02.07 Change in creatinine clearance (ml/min/1.73 m2)

02.08 Change in fractional excretion of sodium (%)

02.09 Change in osmolal clearance (ml/min/100 ml GFR)

02.10 Change in free water clearance (ml/min/100 ml GFR)

02.11 Change in serum Na (mM/L)

02.12 Change in serum K (mM/L)

02.13 Weight change (grams)

02.14 Percent weight change (%)

Contact details for co-reviewers

Dr Deborah Campbell
Director
Section of Neonatology
Children's Hospital at Montefiore
Weiler Hospital, Room 725
1825 Eastchester Road
Bronx
New York USA
10461
Telephone 1: 1 718 904 4105
Facsimile: 1 718 904 2659
E-mail: dcampbel@montefiore.org

This review is published as a Cochrane review in The Cochrane Library, Issue 3, 2007 (see http://www.thecochranelibrary.com for information). Cochrane reviews are regularly updated as new evidence emerges and in response to feedback. The Cochrane Library should be consulted for the most recent version of the review.