In this randomized, controlled non-inferiority trial,
filtered sunlight was compared with conventional phototherapy for the
treatment of hyperbilirubinemia in term and late-preterm neonates. The
primary end point was efficacy, which was defined as a rate of increase
in total serum bilirubin of less than 0.2 mg/dL/h for infants up to 72
hours of age or a decrease in total serum bilirubin for infants older
than 72 hours of age who received at least 5 hours of phototherapy.
Authors pre-specified a non-inferiority margin of 10% for the difference
in efficacy rates between groups. The need for an exchange transfusion
was a secondary end point. Safety, which was defined as the absence of
the need to withdraw therapy because of hyperthermia, hypothermia,
dehydration, or sunburn, was also assessed. Of the total 447 infants,
224 were randomly assigned to filtered sunlight and 223 to conventional
phototherapy. Filtered sunlight was efficacious on 93% of treatment days
that could be evaluated, as compared with 90% for conventional
phototherapy, and had a higher mean level of irradiance (40 vs.
17 µW/cm2/nm; P<0.001). No infant met the
criteria for withdrawal from the study for reasons of safety or required
an exchange transfusion. Authors concluded that filtered sunlight was
non-inferior to conventional phototherapy for the treatment of neonatal
hyperbilirubinemia.
Commentaries
Evidence-based Medicine Viewpoint
Relevance: Phototherapy using blue light is
recognized as a standard of care for the management of majority of
neonates with jaundice, and hence is widely practiced around the globe.
In addition to the wavelength of light used, other issues for effective
phototherapy include mean (and not only peak) irradiance, rate of
irradiance decay, and adequacy of exposure of infants’ body surface
[1,2].
This publication [3] reports a non-inferiority
randomized controlled trial (RCT) comparing filtered sunlight
(Intervention) versus conventional photo-therapy (Comparison) in
infants with mild to moderate neonatal hyperbilirubinemia (Population)
in terms of efficacy and safety (Outcomes). The authors justified this
trial on the basis of individual and societal consequences of
inadequately treated neonatal hyperbilirubinemia, practical difficulties
with delivering appropriate phototherapy in resource-limited settings,
and their previous research experiments with filtered sunlight. The
potential limitations with using conventional phototherapy include (non)
availability of appropriate units, cost of individual units, requirement
of uninterrupted electricity supply, and need to measure irradiance
periodically to ensure efficacious delivery. These can be challenging in
many resource-limited settings.
Prior to the RCT, the investigators demonstrated the
capability of various glass-tinting film materials to selectively
transmit light with wavelength 400 to 520 nm (blue light) while
restricting ultraviolet A and infrared wavelengths [4]. They then
published a protocol [5] for comparing filtered sunlight versus
conventional photo-therapy in near-term neonates <2-week-old in a
hospital setting in Lagos (Nigeria), using a non-inferiority RCT design.
Results from an observational study [6] suggested that filtered sunlight
was safe and efficacious in such infants, although about one-third of
infants required brief periods of omission on account of hyper- or
hypothermia. Simultaneously, they also published data from a survey of
mothers of infants treated with filtered sunlight and reported high
levels of maternal satisfaction with this modality [7]. In a sense, the
preliminary work is important for understanding and appraising this RCT
[3].
Critical appraisal: Table I
highlights the methodological characteristics of the trial using the
standard Cochrane Risk of Bias tool [8]. Overall, the trial can be
considered to have a high risk of bias. This is mostly because of
unclear allocation sequence generation and absence of blinding. The
latter could have been relatively easy for the efficacy component by
ensuring that personnel measuring serum bilirubin were blinded to the
allocation assignment. Of course, it would be more complicated to blind
the outcome assessors for the safety component. It should also be
mentioned that the previous experiments by the investigators [4,6,7]
pointed towards a potentially beneficial effect of filtered sunlight.
This could be expected to bias the investigators towards the inter-vention;
therefore blinded outcome assessments are even more important.
Table I Methodological Appraisal of the Trial
Domain |
Description |
Judgment |
Sequence generation |
Block randomization method with variable block sizes (2-10) was
used. However the specific method used to generate the sequence
was not described. |
Unclear |
Allocation concealment |
Allocation was made using serially numbered slips placed in
opaque sealed envelopes. These were prepared off-site and opened
in sequence after randomization. |
Adequate |
Blinding |
There was no blinding for any of the measured outcomes viz
efficacy (measured by trends in serum bilirubin and/or need for
exchange transfusion) or safety (development of hyper- or
hypothermia, dehydration or sunburn). |
Inadequate |
Incomplete outcome data |
A total of 447 infants were randomized; 224 in the intervention
arm and 223 in the comparison arm. 14 participants did not
receive the planned treatment; 11 in the intervention arm and 3
in the comparison arm. Thereafter 18 and 13 infants respectively
could not be evaluated for efficacy suggesting a total attrition
(from randomization) of 13% in the intervention arm and 7% in
the comparison arm, although these are not the numbers shown in
the flow chart. Although reasons for attrition are clearly
described and appear uncontrollable, the total number of days of
treatment (available for analysis) exactly matched with the
number planned a priori. This appears to be a remarkable
coincidence. The number of randomized infants included in the
safety analysis is unclear. Data analysis is reported to be
intention-to-treat. |
Unclear |
Selective outcome reporting |
All outcomes planned a priori have been reported |
Adequate |
Other sources of bias |
No obvious sources of bias are evident. |
Adequate |
Overall assessment |
|
High risk of
bias |
The investigators incorporated some methodological
refinements in their study. First, this RCT was designed as a
non-inferiority trial despite availability of in-vitro evidence
suggesting superiority of sunlight [9]. The investigators’ own
observational study showed very high efficacy (92%) with filtered
sunlight. These observations could have prompted them to opt for a
superiority trial (i.e proving that filtered sunlight is superior
to conventional phototherapy). In such a study design, a non-significant
difference in efficacy does not automatically mean that the two arms
have equivalent efficacy. Therefore, in such a setting where an
effective treatment (phototherapy) is the standard of care, a
non-inferiority trial (where the design is to show that the intervention
is not much worse than the comparison), is appropriate. To achieve this,
it is required to demonstrate that the 95% CI of the treatment effects
in the two arms lie within a pre-specified narrow range (10% for
efficacy and 5% for safety in this trial). This generally necessitates a
larger sample size than conventional superiority trials. In fact, the
investigators planned to recruit 924 infants based on these
considerations, but chose to work with the ‘number of treatment days’
rather than ‘number of infants’, thereby reducing the sample size
requirement to less than half.
Unfortunately the reduced sample size made it
feasible to complete the trial within 10 months (November 2012 to
September 2013) making it impossible to study the efficacy during two
months of the year. In general, the average highest and lowest
temperature in Lagos during the missing months (September and October)
are not very different from the other months [10]; however these are two
of the wettest months, with precipitation during at least one-third of
the days. Naturally, a twelve month period of observation would have
given a better understanding of the effect of ambient climatic
conditions.
Despite high overall efficacy (over 90% as per the
investigators’ definition), the rate of bilirubin decline in the
conventional phototherapy arm appears to be somewhat lower than in other
trials of phototherapy [11-13]. Of course, it may not be appropriate to
compare efficacy across trials unless there is homogeneity in terms of
population, intervention and timing of outcome measurement. The
importance of this is that a less-than-adequate rate of bilirubin
decline in the phototherapy arm would create a spurious impression of
greater efficacy in the filtered sunlight arm, as the authors have
reported [3]. This ‘spurious’ superiority is itself doubtful because
there was overlap between the 95% confidence intervals, suggesting
comparable effects.
Although the investigators reported comparable safety
in the two arms of the trial [3] for most outcomes measured, there was
greater incidence of hyper- and hypothermic episodes in infants
receiving filtered sunlight. The comparability was achieved through
stringent hourly monitoring of body temperature, and immediate
institution of counter measures. This suggests that routine clinical use
of filtered sunlight would necessitate stringent monitoring for adverse
effects and facilities to correct these. This may entail greater
manpower and/or resource usage, negating the putative benefits of
availability at the point-of-care and lower cost (of the equipment) with
filtered sunlight.
The comparable absence of withdrawals (from both
groups) on account of need for exchange transfusion attests to the
inclusion of relatively milder cases and lower thresholds for treatment.
It is unclear whether the intervention would perform similarly in
real-world conditions.
A finding that was not sufficiently emphasized is
that one-seventh of infants receiving filtered sunlight required
phototherapy after daylight hours [3]. This occurred despite initiating
treatment at a lower threshold serum bilirubin than the standard
guideline of the American Academy of Pediatrics [3,14]. This has two
important implications. First, it is likely that treating at the usual
thresholds would necessitate continuing the intervention in a larger
proportion of infants. Second, the observation mandates backup
phototherapy units thereby limiting its usage to health-care facilities
with adequate manpower and resources to manage sick neonates. This
practically neutralizes the proposed advantages of filtered sunlight
described above. A minor but potentially important point is the
stringency with which irradiance was measured in this trial. The
intention was to measure irradiance once daily in the phototherapy arm
and hourly in the filtered sunlight arm. If this was followed, we would
expect 325 readings in the phototherapy arm, whereas only 293 are
reported. We would also expect 1876 readings in the filtered sunlight
arm (maximum of 7 readings per day × 268 days) whereas 2959 are
reported. These deviations from the protocol have not been explained.
How would mothers respond to offering their babies
filtered sunlight as opposed to conventional phototherapy? On the one
hand, sunlight exposure was a widely practiced traditional remedy for
newborn jaundice; on the other hand, this has been actively discouraged
on account of potential harmful effects of ultraviolet and infra-red
radiation. The authors’ previous study [7] suggested a high level of
maternal satisfaction with filtered sunlight, using a five point Likert
scale. However in the survey, mothers reported a median score of at
least 4 for all issues on which feedback was sought (filtered sunlight
was only one of these) suggesting high level of satisfaction with the
overall health-care delivery system in general, rather than filtered
sunlight in particular.
Extendibility: It is clear that usage of
filtered sunlight for neonatal jaundice can be considered only in
infants with relatively mild jaundice and with intensive monitoring, as
well as the presence of backup phototherapy units. This severely
restricts the potential of using it as a primary health-care delivery
measure in rural areas and at the point-of-care.
Some of the considerations that could make it easy to
extrapolate the findings of this trial [3] include abundance of ambient
sunshine in most parts of India, during most of the year; and the
potential for constructing filtered sunlight tents at low cost. The
issues that could preclude direct extrapolation include the geographic
(latitude) location of Lagos and the relatively uniform high skin
phototype score of participants in the trial. It is also unclear how the
irradiance would vary with environmental air pollution and smog/fog in
urban settings in Northern India.
Conclusions: This RCT suggests that filtered
sunlight could be non-inferior to conventional phototherapy for mild to
moderate neonatal jaundice in terms of efficacy and safety, in the
presence of adequate facilities for backup phototherapy and intensive
monitoring. These effects were evident at an irradiance level more than
twice that of conventional phototherapy. There are several
methodological and feasibility considerations that preclude the
immediate application of filtered sunlight as a viable alternative to
conventional phototherapy.
References
1. Bhutani VK. Committee on Fetus and Newborn;
American Academy of Pediatrics. Phototherapy to prevent severe
neonatal hyperbilirubinemia in the newborn infant 35 or more weeks
of gestation. Pediatrics. 2011;128:e1046-52.
2. Subramanian S, Sankar MJ, Deorari AK,
Velpandian T, Kannan P, Prakash GV, et al. Evaluation of
phototherapy devices used for neonatal hyperbilirubinemia. Indian
Pediatr. 2011;48:689-96.
3. Slusher TM, Olusanya BO, Vreman HJ, Brearley
AM, Vaucher YE, Lund TC, et al. A randomized trial of
phototherapy with filtered sunlight in African neonates. N Engl J
Med. 2015;373:1115-24
4. Vreman HJ, Slusher TM, Wong RJ, Schulz S,
Olusanya BO, Stevenson DK. Evaluation of window-tinting films for
sunlight phototherapy. J Trop Pediatr. 2013;59:496-501.
5. Slusher TM, Olusanya BO, Vreman HJ, Wong RJ,
Brearley AM, Vaucher YE, et al. Treatment of neonatal
jaundice with filtered sunlight in Nigerian neonates: study protocol
of a non-inferiority, randomized controlled trial. Trials.
2013;14:446.
6. Slusher TM, Vreman HJ, Olusanya BO, Wong RJ,
Brearley AM, Vaucher YE, et al. Safety and efficacy of
filtered sunlight in treatment of jaundice in African neonates.
Pediatrics. 2014;133:e1568-74.
7. Olusanya BO, Imam ZO, Mabogunje CA, Emokpae
AA, Slusher TM. Maternal satisfaction with a novel filtered-sunlight
phototherapy for newborn jaundice in Southwest Nigeria. BMC Pediatr.
2014;14:180.
8. The Cochrane Collaboration’s Tool for
Assessing Risk of Bias. Available from: http://ohg.cochrane.org/sites/ohg.cochrane.org/files/uploads/Risk%20of%20bias%
20assessment%20tool.pdf. Accessed November 15, 2015.
9. Salih FM. Can sunlight replace therapy units
in the treatment of neonatal jaundice? An in vitro study.
Photodermatol Photoimmunol Photomed. 2001;17:272-7.
10. Weather Statistics for Lagos (Nigeria).
Available from: http://www.yr.no/place/Nigeria/Lagos/Lagos/statistics. html.
Accessed November 15, 2015.
11. Brandão DC, Draque CM, Sañudo A, de Gusmão
Filho FA, de Almeida MF. LED versus daylight phototherapy at low
irradiance in newborns
³35
weeks of gestation: randomized controlled trial. J Matern Fetal
Neonatal Med. 2015; 28:1725-30.
12. Mohammadizadeh M, Eliadarani FK, Badiei Z. Is
the light-emitting diode a better light source than fluorescent tube
for phototherapy of neonatal jaundice in preterm infants? Adv Biomed
Res 2012;1:51.
13. Tridente A, De Luca D. Efficacy of
light-emitting diode versus other light sources for treatment of
neonatal hyperbilirubinemia: A systematic review and meta-analysis.
Acta Paediatr. 2012;101:458-65.
14. American Academy of Pediatrics Subcommittee
on Hyperbilirubinemia. Management of hyperbilirubinemia in the
newborn infant 35 or more weeks of gestation. Pediatrics.
2004;114:297-316.
Joseph L Mathew
Department of Pediatrics,
PGIMER, Chandigarh, India.
Email:
[email protected]
Neonatal hyperbilirubinemia is a common problem in
the first few days of life. Phototherapy is the standard treatment for
the management of neonatal hyperbilirubinemia, and most cases can be
successfully controlled by its use. In few cases where phototherapy
fails, exchange transfusion can be employed to rapidly lower serum
bilirubin levels to safe limits. Although phototherapy units are
available in resource-limited settings at affordable costs, the
performance of these units leaves much to be desired due to low
irradiance (energy output) or erratic electric supply. Due to this, many
newborns are exposed to sub-therapeutic irradiance levels.
Exposing jaundiced newborns to sunlight is practiced
in many societies, although health professionals do not advocate its use
due to doubtful efficacy and safety concerns regarding sunburn,
hyperthermia or exposure to ultraviolet radiation [2]. There are also
issues of availability of sunlight round the clock and during cloudy
weather, and its use during winter months when the risk of hypothermia
is high. The present study has refuted some of these concerns and
demonstrated that filtered sunlight is equally effective and safe. The
authors used special film canopies which filtered out most ultraviolet
and some infrared (heat) radiation while allowing passage of blue light.
Unlike conventional phototherapy, where a newborn requires separation
from mother to receive phototherapy, filtered sunlight has the advantage
of maintaining mother-infant interaction with all its attendant
benefits, if a large enough canopy is used to accommodate the
mother-newborn dyad. This is a low cost technology and has other
advantages as well, such as availability in remote regions, use of
energy source which is essentially free, and also eco-friendly. By
virtue of its nature, filtered sunlight cannot be used to treat
hyperbilirubinemia in sick newborns who require close monitoring in a
NICU environment.
Before filtered sunlight can be adopted on a wider
scale to treat neonatal hyperbilirubinemia, studies are needed to assess
its impact in other settings. A limitation of this study is that it
included newborns with serum bilirubin levels below 15 mg/dL. Whether
filtered sunlight will be equally effective at higher bilirubin values
or rapidly rising bilirubin levels needs to be studied further. At high
bilirubin levels, nighttime phototherapy will be required. This occurred
in 13% of newborns in present trial. Thus conventional phototherapy is
not going to disappear. The filtered sunlight will complement
conventional phototherapy rather than replace it. The present trial has
vindicated the long-held community practice of exposing jaundiced
newborns to sunlight, albeit in a modified manner.