A few recent studies have noted the marked variability between sites in the use of medications in preterm infants with BPD.
Nelin TD, et al. The association between diuretic class exposures and enteral electrolyte use in infants developing grade 2 or 3 bronchopulmonary dysplasia in United States children’s hospitals. J Perinatol. 2021.
Bamat NA, et al. Loop Diuretics in Severe Bronchopulmonary Dysplasia: Cumulative Use and Associations with Mortality and Age at Discharge. J Pediatr. 2020.
Greenberg JM, et al. Respiratory medication use in extremely premature (<29 weeks) infants during initial NICU hospitalization: Results from the prematurity and respiratory outcomes program. Pediatr Pulmonol. 2020;55(2):360-8.
Tan C, et al. Diuretic use in infants with developing or established chronic lung disease: A practice looking for evidence. J Paediatr Child Health. 2020;56(8):1189-93.
The first 3 of these studies are multi-institutional observations, the 4th being from a single centre. Variations in medication use and in particular the use of diuretics are enormous, some centres seem to treat almost all of their babies with loop diuretics, while in others such use is limited. In my centre chronic diuretic use is quite unusual, while intermittent brief courses may be tried in a small percentage of babies.
Why such variation? One of the usual explanations for such variation is an absence of a strong evidence base; so what is the evidence for diuretic use for the prevention or treatment of BPD?
Diuretics were first tried I think because the early phases of BPD show interstitial oedema; so why not use a diuretic? The few mechanistic studies that were performed showed an improvement in some measures of lung mechanics, but little or no effect on gas exchange. One controlled trial of prolonged use in infants with BPD on non-invasive O2 therapy showed some decrease in FiO2, but no decrease in the duration of O2 therapy, and no residual effect after the furosemide was stopped.
Interestingly lung mechanics changes have also been shown in anephric dogs, the ion pump which is inhibited by the loop diuretics (NaK2CL co-transport) is present on the luminal surface of the thick ascending limb of the Loop of Henle. furosemide also inhibits chloride transport in the lungs, which may be relevant to its effects, especially in dogs without kidneys, the mechanism is probably related to Cl Na co-transport. There is a disconnect between diuresis and effects on pulmonary mechanics, for example, inhaled furosemide can improve compliance and resistance in infants with BPD, but without causing a diuresis. Another placebo-controlled study showed effective improvement in lung mechanics with alternate day furosemide, without an overall increase in urine output.
Improving mechanics of course is not really what I want when I prescribe a medication, I want to improve clinical status and clinical outcomes. There is little or no evidence that diuretics of any kind do this. The Cochrane review of loop diuretics in BPD found 6 trials (all furosemide) and no evidence of impact on clinically important outcomes (which weren’t reported in most trials). The Cochrane review of thiazides in BPD (both reviews are authored by Luc Brion, and were last updated in 2011, but I am not aware of any new trials) also showed in six trials no clinically significant effect. The only exception being the RCT of thiazides in intubated babies with BPD which showed less mortality with thiazides, but that trial only included 34 babies, the 15 controls were all boys except one, and with a higher pip and mean airway pressure than the 19 diuretic babies (8 females), therefore a significant chance of a type 1 error.
As a reminder to everyone, the Neonatal Cochrane reviews are all available at the Vermont Oxford Webisite, free of charge for anyone. https://public.vtoxford.org/cochrane-at-von/
What do these new publications say? The first, Nelin et al shows that babies on any diuretic also often receive mineral supplements and that babies on thiazides were more likely to receive them than babies on loop diuretics. That to me is entirely expected, thiazides often lead to hyponatraemia, whereas it is less frequent with furosemide. Chronic thiazide use often is accompanied by chronic sodium supplementation, which is a questionable combination, as you are using the diuretics because they cause natriuresis! Pushing sodium out by the kidneys and simultaneously supplementing it enterally makes little sense. Loop diuretics, however, cause more of a reduction in total body fluid, so less dilution of serum sodium. The study also calls into question the concurrent use of potassium-sparing diuretics, as babies receiving them in addition to thiazides (the most common use in this trial) were just as likely to receive potassium supplements. Again that should be of no surprise, spironolactone does not affect thiazide-induced potassium loss in the newborn infant. An RCT from 2000 showed that adding spironolactone to a thiazide had no additional benefit on lung mechanics, and did NOT spare potassium. Just as many babies received K supplements, and serum potassium was identical with and without spironolactone. As far as I can see, that study is the only scientific data that exists about the value of adding spironolactone to a thiazide, i.e. there is no value.
The second publication, Bamat et al showed, not for the first time, that loop diuretic use is extremely variable, and that the variability is explained by which hospital you are in, rather than the severity of your lung disease. The study examined data from infants with grade 2 or 3 BPD and showed that infants from hospitals that used a lot of loop diuretics did not get home any sooner, (PMA at discharge was around 47 weeks, showing these were quite sick babies) but were more likely to go home on diuretics. The hospitals ranged from use of loop diuretics on 8% of the days between birth and discharge among infants with moderate and severe BPD to 50%.
The third in my list , Greenberg et al is a prospective multicentre cohort of infants born between 23 and <29 weeks gestation, which produced this pretty figure :
Of note, this study was not solely babies with BPD, but nevertheless by 5 weeks of age half of the babies are on furosemide. Also, from the supplemental data, 45% of the babies who did not develop BPD had received furosemide before they reached 40 weeks.
The last of these studies (Tan et al) is a single centre publication from Monash and deals mostly with the thiazide/spironolactone combination. As a single centre study they were able to give some more data pre and post diuretics, which showed that 84% became hyponatraemic and 12% hypokalaemic after diuretics, which did not significantly improve gas exchange but did lead to a slow down in weight gain.
None of this is new. Slaughter JL, et al. Variation in the use of diuretic therapy for infants with bronchopulmonary dysplasia. Pediatrics. 2013;131(4):716-23.Laughon MM, et al. Diuretic exposure in premature infants from 1997 to 2011. Am J Perinatol. 2015;32(1):49-56. Our use of these medications, in particular, the diuretics, is non-evidence-based, extremely variable, off-label, and somewhat irrational.
We should do the following:
- Stop all use of spironolactone in the newborn. It does not improve electrolyte status, does not spare potassium, does not add to diuretic effects of thiazides, and has adverse effects, in particular blocking androgen receptors, but also with reports of thrombocytopaenia and agranulocytosis.
- Stop routine use of diuretics to prevent or treat BPD. There is no evidence that diuretics reduce the incidence of BPD, and no evidence of clinically important benefits in treatment.
- Stop prolonged use of diuretics to prevent or treat BPD. Prolonged use leads to serious electrolyte disturbance, not to mention nephrocalcinosis and bone demineralization, metabolic alkalosis and chloride depletion.
- Perform prospective controlled studies to determine potential indications for diuretic use and the balance of risks and benefits. The lack of clinical impact when comparing widespread to restricted diuretic use in these observational studies is clear evidence that equipoise should exist, and clinical trials are essential for these widely used agents.