The scenario is often the same. Faced with a child born to a mother with risk factors for sepsis you decide to start antibiotics. The time comes closer to 36 – 48 hours when you must decide whether or not to continue. Each time we examine our results and look at cultures and try to do what is right. Yet defining right is sometimes hard for so many. If we had 100% sensitivity and specificity for all our tests it would be easy but we don’t. So what can we do?
If I had to have one wish though it would be that we could improve upon our diagnostic accuracy when it comes to treating suspected infections in the newborn. As health care providers we have an extremely loud inner voice trying to tell us to minimize risk when it comes to missing a true bacterial infection. On the other hand so much evidence has come forth in the last few years demonstrating that prolonging antibiotics beyond 48 hours is not just unwise in the absence of true infection but can be dangerous. Increased rates of necrotizing enterocolitis is just one such example but other concerns due to interfering with the newborn microbiome have arisen in more recent years. What follows are some general thoughts on septic workups that may help you (and myself in my own practice) as we move ahead into the New Year and may we cause less harm if we consider these.
The Role of Paired Blood Cultures
Although not published by our centre yet, we adopted this strategy for late onset sepsis a couple years back and have seen a significant reduction in work-ups deemed as true infections since adoption. While the temptation to do only one blood culture is strong as we have a desire to minimize skin breaks consider how many more there will be if you do one culture and get a CONS organism back. There will be several IV starts, perhaps a central line, repeat cultures etc. If you had done two at the start and one was positive and the other negative you could avoid the whole mess as it was a contaminant from the start. On my list of do no harms I think this may have the greatest benefit.
The Chest X-Ray Can Be Your Friend
While I am not a fan of routine chest x-rays I do believe that if you are prepared to diagnose an opacification on a chest x-ray as being due to a pneumonia (VAP or in those non-ventilated) that you need to follow this up with a repeat x-ray 24 – 48 hours later. If the opacity is gone it was atelectasis as a true pneumonia will not clear that easily. Well worth the radiation exposure I say.
If You Are Going To Do a Work-up Make It A Complete One
We hear often in rounds the morning after a septic work-up that the baby was too sick to have an LP and that we can just check the CSF if the blood is positive. There are two significant problems to this approach. The first which is a significant concern is that in a recent study of patients with GBS meningitis, 20% of those who had GBS in the CSF had a negative blood culture. Think about that one clearly… relying on a positive culture to decide to continue antibiotics may lead to partially treated GBS meningitis when you discontinue the antibiotics prematurely. Not a good thing. The second issue is that infants with true meningitis can have relatively low CSF WBC counts and may drift lower with treatment. Garges et al in a review of 95 neonates with true meninigits found that CSF WBC counts >21 cells per mm3 had a sensitivity of 79% and specificity at 81%. This means that in those with true meningitis 19% of the time the WBC counts would be below 21 leading to the false impression that the CSF was “fine”. If antibiotics were effective it could well be by 48 hours that the negative CSF culture you find would incorrectly lead you to stop antibiotics. Message: Do the CSF sampling at the time of the septic work-up whenever possible.
If We Aren’t Prepared To Do a Supra Pubic Aspirate Should We Not Collect Urine At All?
This provocative question was asked by a colleague last week and is based on the results of a study which was the topic of the following post: Bladder Catherterizations for UTI: Causing more harm than good? The gist of it is that it would appear that in many cases the results of a catheter obtained urine cannot be trusted. If that is the case then are we ultimately treating infections that don’t actually exist when the only positive culture is from a urine. I believe using point of care ultrasound to obtain specimens from a SPA will be the way to go but in the meantime how do we address the question of whether a UTI is present or not? May need to rely on markers of inflammation such as a CRP or procalcitonin but that is not 100% sensitive or specific either but may be the best we have at the moment to determine how to interpret such situations.
Lastly, Slow Down And Practice Good Hand Hygiene
So much of what I said above is important when determining if an infection is present or not. The importance of preventing infection cannot be understated. Audits of hand hygiene practice more often than not demonstrate that physicians are a group with some of the lowest rates of compliance. Why is that? As a physician I think it has nothing to do with ignorance about how to properly perform the procedure but rather a tendency to rush from patient to patient in order to get all the things done that one needs to do well on service or call. If we all just slow down a little we may eventually have less need to run from patient to patient as the rate of infections may drop and with it demand for our time.
If slowing down is something that you too think is a good idea you may want to have a look at the book In Praise of Slowness by Carl Honore (TED Talk by Carl Below) which may offer some guidance how to do something that is more easily said than done. Here is hoping for a little slower pace in the new year. We could reap some fairly large benefits!
In summary though, the trial involved 118 infants who received 40% dextrose gel vs 119 who received a placebo gel. All of the infants in this study were selected based on risk factors for hypoglycemia (IDM, IUGR, LBW, LGA, near term) and were all 35 weeks or greater. Each infant had to be less than 48 hours of age when enrolled. Infants received 0.5 mL/kg 40% dextrose gel (200 mg/kg). This was designed to deliver the same amount of sugar as would be given with a D10W bolus of 2 mL/kg. In order to receive the treatment the blood glucose had to be < 2.6 mmol/L so equivalent to our own standards in Canada and the US. Treatment failure, which was the primary outcome was defined as a blood glucose < 2.6 mmol/L despite two treatments with gel. The studies most important findings were a reduction in NICU admission and greater breastfeeding rates at 2 weeks of age (due to avoidance of formula feeding to keep the glucose stable).
But Is It Safe?
With any new strategy though, questions arise regarding safety of the product in the long term. As one reader commented after the original post, could the red dye be harmful in some way or perhaps some other constituent of the gel? The authors of the original study have now published the follow-up paper entitled Outcome at 2 Years after Dextrose Gel Treatment for Neonatal Hypoglycemia: Follow-Up of a Randomized Trial. The findings were a little concerning to me in that there was a high rate of neurosensory impairment in both arms (sham and glucose gel) with the following findings. None of the differences were significant however.
Looking at these results you could be dismayed that the glucose gel did not show any benefit compared to standard therapies for hypoglycemia but if you look at the original study one could equally ask why would we have expected it to?
Should The Results Surprise Us?
All of the newborns in either arm had an episode of hypoglycemia recorded in order to qualify for entry into the study. The glucose gel was effective compared to placebo in reducing admissions and increasing breastfeeding rates (which one might think would improve outcomes) but we don’t know what the drop off rate in breastfeeding was after 2 weeks. What I can say though is that if there were significant adverse side effects with the glucose gel I would have expected to see some differences in outcome favouring the placebo group which did not occur.
An additional issue is we know that the placebo group had more treatment failures meaning they would have had more newborns in the original study with a second low glucose. Would this favour a worse outcome in the placebo group? If so does the equivalence in groups suggest that the dextrose gel might worsen outcomes?
Unfortunately, what this study is really missing is some indication of how low and how long the blood sugars remained under 2.6 mmol/L in both arms of the study. We know the mean low glucoses were similar but what about duration and range? While the rates of mild, moderate and severe impairment are the same we don’t know how severe the hypoglycemia was which if unbalanced between the groups could actually lead to very different conclusions here. For example, let’s say the glucose gel group had an over representation of infants with 3 or more episodes of hypoglycemia compared to the placebo arm. The fact that the outcomes are equivalent would suggest that the glucose gel is in fact protective.
What Can We Say?
I suspect that while glucose gel is effective, to truly assess harm across many different aspects of development we will need larger sample sizes. We also have to take the results of this study with a grain of salt as so many that have come before it have seen outcomes at school age reveal different findings than when assessed at 2 years of age as in this study. From my standpoint though I will continue to advocate for the use of glucose gel as the reduction in NICU admissions and enhancement of breastfeeding rates especially if sustained are well worth the efforts to implement this strategy if you aren’t using it already.
I will admit it. I resist change at times just like many others. This may come as a surprise to some of you who have worked with me and accused me of bringing too much change at times to the units. The truth though is that when one understands something and is enthusiastic about implementation the change does not seem so difficult. When it isn’t your idea though we may find ourselves a little uneasy about adopting this unfamiliar practice.
Such has been my experience with nasal HFOV. It is a strategy that has been around for over five years but has seen slow adoption among centres in Canada and has trickled into practice in Winnipeg on a few occasions. In each occasion when I have been asked about either continuing or perhaps starting this therapy I have shrugged my shoulders and confessed my inexperience with the modality. Sure I have used HFOV through an ETT but through prongs or a mask?! How would it work? Could it cause harm? What would the actual indications be? How would our in house physicians and NNPs respond to abnormal gases overnight even if I felt comfortable with using it? These sorts of questions have led to virtual inertia in my acceptance of the strategy.
Before I go on it would be good to see an example of how it is set up. The MedinCNO device is capable of delivering such non-invasive HFOV and can be seen in this short video.
One could use the strategy either prophylactically to extubate an infant or as rescue to prevent reintubation if trials of either CPAP or NIPPV were unsuccessful. HFOV is known to be very effective at clearing CO2 when used through an ETT so perhaps nasal application could also lower pCO2 and achieve a similar effect. This was tested using a neonatal lung simulator by Mukeji A et al Nasal high-frequency oscillation for lung carbon dioxide clearance in the newborn. In this study CO2 was introduced into the manequin and the amount of exhaled CO2 determined while on CPAP, NIPPV and nasal HFOV. Interestingly during CPAP no exhaled CO2 could be detected while CO2 clearance occurred during NIPPV and nasal HFOV although it was three-fold greater with HFOV. In theory then CO2 clearance would appear to be better so in the case of ventilatory failure as evidenced by CO2 retention this modality would seem to win out.
Clinical Evidence for Use
There is one RCT in term infants with TTN to support the practice while the rest are unblinded case series with no controls. Four Canadian NICUs recently described their experience however using a retrospective analysis. Included were 79 instances of HFOV distributed as follows; 73% utilized as rescue from another mode and in 27% used as the primary mode for extubation. The outcomes are shown in the table:
In 45% of cases the patients needed intubation after first trialing CPAP or NIPPV while in 33% of cases following extubation the infants needed replacement of the endotracheal tube. The numbers here are small so it is difficult to truly compare them to other studies with confidence but reintubation rates of 40-44% have been noted recently when using NIPPV or CPAP so the numbers are at least consistent.
One aspect though that caught my eye was the duration of use for HFOV across these 79 patients. The median use was 57 hours with the longest duration being just over 400 hours. It would seem that the use of this modality for the most part is as a bridge to something else. The median duration of 2.5 days is much shorter than the weeks that some of our smallest infants remain on CPAP/NIPPV for. Whether for rescue or prophylaxis this is not a long term option.
Another point worth noting though is the question of whether it is the pressure or oscillatory wave that is leading to success. As the authors note, there were a wide range in applications of MAP, delta P and frequency.
MAP ranges from 8 – 24 cm H2O while frequency from 6 – 14 hz and amplitude varied widely depending on the device used but was as high as 100%. While high MAP has been used invasively though an ETT I can’t help but wonder if in some cases the real benefit was the high MAP. What would happen for example if the centres had simply raised the CPAP to 10, 12 or even higher?
In the end it would seem that in principal it is an effective therapy that may be able to remove CO2 more efficiently than the other modes. What we don’t have are RCTs in the smallest babies comparing HFOV to NIPPV or CPAP with adequate power to detect differences. I suspect these will come soon enough but what do we do in the meantime? The main reservation I have has to do with safety. We truly have little if any data on this without proper trials to ease such worry. When a patient is in front of us though and is failing CPAP or NIPPV what are we to do? Should we intubate or trial this modality based on the evidence thus far?
I might be tempted to trial HFOV in this circumstance but as with any new therapy we need education for all staff. Everyone caring for our infants need to understand what they are using and how to respond based on clinical findings. This is the real issue with safety that I see and until such time that we have widespread education across RRT, nursing and medicine I would suggest we use this with trepidation. This is not a rejection of the modality in the least but rather a call to come together as a team and see how implement this in such a way that will provide direction to caregivers, provide a consistent approach with respect to length of use, indications and when to change direction entirely. Time to call a meeting of the minds I think.
One of my very early posts on this blog pertained to my fascination with an Israeli strategy of monitoring end tidal CO2 in place of drawing blood gases. Please see A Strategy to Minimize Blood Sampling in ventilated premature and term infants. The gist of this strategy is that by sampling distal CO2 measurements near the carina you obtain a non diluted sample of CO2 as compared to the traditional proximal end tidal measurement. The authors have shown this to be highly accurate compared to comparable arterial samples during both conventional and high frequency oscilatory ventilation.
This month the same group published a follow-up study that has me excited again about this technological approach. The paper is entitled Impact of Continuous Capnography in Ventilated Neonates: A Randomized, Multicenter Study. In this multicentre study 55 infants were ventilated through double lumen endotracheal tubes. The first group of 25 were randomized to an open group in which the continuous CO2 measurement was available to clinicians to base ventilator management on. The second group had these values recorded but the treating physicians were blind and based ventilatory changes on arterial blood gas sampling. The primary outcome was based on the percentage of time pCO2 was kept between 30 – 60 with the goal range for permissive hypercapnea being 45 – 55.
“Compared with infants in the masked group, those in the monitored group had significantly (P = .03) less time with an unsafe dETCO2 level (high: 3.8% vs 8.8% or low: 3.8% vs 8.9%). The prevalence of intraventricular hemorrhage or periventricular leukomalacia rate was lower in the monitored group (P = .02) and was significantly (P < .05) associated with the independent factors dETCO2 monitoring and gestational age.”
The same held true for the blood gas values with statistically significant benefits in terms of keeping CO2 within the predefined safe range.
The findings make a great deal of sense. Sampling continuously and adjusting ventilation in real time (assuming a strong correlation between EtCO2 and blood gases) makes sense. Blood gases done even every four hours can not compare to having results available continuously. As such I tend to believe the findings of better accuracy. Although the incidence of IVH and PVL are part of the secondary outcomes there is significant biologic plausability to these findings. We know that cerebral blood flow is very sensitive to PCO2 with low values causing cerebral vasoconstriction and high values dilation of these same vessels. Furthermore having wide swings in CO2 could lead to periods of ischemia and then significant reperfusion with resultant injury. Finding that these two outcomes are increased with intermittent sampling therefore is plausible and deserving of further study.
Challenges That Need to Be Addressed
Since the publications by the same group showing the accuracy of this method of sampling CO2 we have tried this strategy for several infants. In short, it works and is very accurate. We have been able to avoid many blood gases along the way but there remains a significant challenge with avoiding blockage of the second port for sampling. Furthermore, secretions if getting past the end of the port and into the microstream sampling device can damage the analyzer such that we have only one left functioning in our units.
As eager as I am to roll this out as a change in practice we need to address these technical issues but once accomplished this may truly have an impact as a cerebral protective strategy for the preterm infant and with time and further long term outcomes available offer a meaningful way of reducing disability in this population.
Before we get into critiquing such a study it is worth explaining what laser acupuncture is. From the website acupuncture today I found the following background:
“In 1991, a study was done in Novosibirsk, Russia that applied directly to the study of acupuncture. Researchers shined light on various parts of the body and found that light traveled under the skin to other acupuncture points, but it didn’t travel to places that were not on acupuncture meridians. It appears that the body contains a sort of fiber optic network—where light enters an acupuncture point, travels through the meridian and can be detected at other places along the meridian with a sensitive photon detector. This is a fascinating study showing how light is actually received, used and transmitted throughout the body.”
This study compared patients receiving treatment for NAS with morphine and phenobarbital and compared 14 in each group on a blinded fashion to laser acupuncture or nothing in addition to the standard regimen. They avoided the need for a placebo in that the intervention was done by the health care provider and then a separate person calculated a pain score who was blind to whether the laser had been provided or not. By blinding the people scoring they hoped to address the criticism that would be present had the study been unblinded. The acupuncture was performed daily at five ear and four body points bilaterally with the primary outcome being duration of oral opiate use. The authors found a reduction in days of opiate treatment but being a small study it would be considered a pilot at best. The results were 28 days (22 to 33) vs 39 days (32 to 48), respectively, P =.019 and the authors make a comment that phenobarb levels were similar between the groups as well but with overall shorter lengths of stay of 35 days (25 to 47) vs 50 days (36 to 56), respectively, P = .048. There are several issues with the study though and they are as follows. Yes, the light therapy was provided behind a closed door but how do we know that the providers truly kept it a secret during the whole stay of the patient? What medications were the mothers taking and at what doses and for what duration? Was the exposure between groups really the same? Also there was no standard means of weaning that I can see so how do we know by chance that the difference is simply who was on at the time with respect to physician deciding on aggressiveness of weaning. Another issue is that women on replacement medications such as methadone or suboxone were excluded as were women with polysubstance abuse. How close to our patient population really were these patients? Lastly, nearly significant was a preponderance of males in the conventional group and we all know how males generally do compared to females in most studies. Nonetheless it is a exciting study to imagine an effect from and that is what the authors I think were banking on.
What about laser acupuncture for treatment of pain?
One other study comparing sucrose to laser acupuncture for prevention of pain with heel lances. In this study of 42 infants randomized to the intervention or sucrose the authors concluded that sucrose was superior to acupuncture for managing pain from heel lancing. This study did not have a third arm in which no treatment was given (likely due to it being unethical) so what we can’t say for sure is whether laser is better than nothing.
Personally I find a bit of a leap to believe that shining a light through the skin would have an effect on pain perception by anyone. I can certainly understand the concept of putting a needle into tissue or applying pressure no different from the effect of rubbing ones temples when they have a headache (works for me at least). The Chemistry major in me though has trouble with the effect of photons passing through tissue even if directed at acupuncture points. Having said that the only positive thing I can say is that a quick search for such instruments shows them to be commercially available and as cheap as $50 or so. I have no doubt there are people who will read this RCT and be impressed with its additional purported impact on NAS and thankfully if they do decide to pursue this intervention they will not have spent a fortune on it. Money in health care should be directed to what works to help patients and the skeptic in me just isn’t buying this even if it appeals to my curiosity with non-invasive strategies.