Liquid Crystal Thermometry for the Detection of Neonatal
Hypothermia in Nepal
by N. Manandhar,* M. Ellis,** D. S. Manandhar,* D. Morley,** and A. M. de L. Costello**
* Prasuti Griha Maternity Hospital, Kathmandu, Nepal
**Centrefor International Child Health, Institute of Child Health, London, UK
Summary
We assessed the sensitivity, specificity and likelihood ratio of a low cost liquid crystal strip
thermometer (LCT) compared with axillary mercury thermometry for the detection of neonatal
hypothermia in Nepal. The subjects were 76 healthy newborns in the government maternity hospital
of Kathmandu, Nepal in winter.
The validity of LCT for the detection of neonatal hypothermia (less than 36°C) showed a sensitivity
of 83 per cent, specificity 96 per cent, positive predictive value 98 per cent and a likelihood ratio of 23.
Use of LCT on newborns in this setting raises a measured pretest probability of first day hypothermia
of 63 per cent to a post-test probability of 97 per cent
Liquid crystal thermometry is a simple, low-cost, and valid method for identifying core
hypothermia in newborns. It is ideal for isolated rural communities where LCT strips could be added
to delivery kits.
Introduction
Neonatal hypothermia is a common problem in many
developing countries.'" Previous studies in Prasuti
Griha Maternity Hospital, Kathmandu have documented
a high prevalence of neonatal hypothermia.4"5 Hypothermia is associated with an increased risk of mortality
in a developing country hospital setting.
If temperature of newborn infants is measured in
developing countries, a mercury-in-glass thermometer is
generally used. Rectal temperature measurement carries
a small risk of bowel perforation7 and cross-infection.8
Several studies have shown a close correlation between
rectal and axillary temperatures using this method.910
Axillary temperature measurements are recommended
by the American Academy of Pediatrics. More recently,
the World Health Organization has recommended the
use of low-reading thermometers in newborn care to
detect hypothermia below 35°C."
Unfortunately, mercury-in-glass thermometers, especially the low-reading variety, are fragile and difficult to
obtain in many parts of the developing world.
This study evaluated the use of a low cost liquid
crystal thermometer (LCT) for the detection of hypothermia in newborn infants. A previous study of 498
Acknowledgements
We are grateful to the British Overseas Development Administration for their financial support. Dr M. Ellis is a research
fellow funded by the Wellcome Trust We thank the staff of the
MIRA project for their continuing support.
Correspondence: Dr Anthony Costello, Institute of Child
Health, 30 v-.ilford St, London WC1N 1EH, UK
Journal of Tropical Pediatrics
Vol.44
February 1998
under-3-year-olds in Africa showed that LCT reads
lower than rectal mercury thermometers by around
1.7°C.12 The authors concluded that although LCTs
underestimated fever in children the method was useful
for detecting hypothermia (sensitivity 100 per cent,
specificity 92 per cent) in that population. No formal
evaluation of LCT for the detection of neonatal
hypothermia has been reported. We report an evaluation
of the use of LCT in a Nepali maternity hospital in winter
when hypothermia is common.
Methods
Location
The study was conducted at Prasuti Griha, a large
government-funded maternity hospital in Kathmandu,
Nepal. The hospital, the largest maternity unit in the
country, has 250 beds, and delivers 14000 infants
annually. In winter, ambient temperatures on the
postnatal wards are rarely above the WHO recommended minimum of 25°C, and postnatal hypothermia
and cold stress are common among healthy newborns.
Subjects
Healthy, term, newborn infants (n = 76) on the first day
of life were recruited by incident density sampling on the
postnatal wards. Infants were enrolled after the study
was explained to their mothers and their verbal
permission sought.
Study protocol
After enrolment, one investigator (NM) placed the LCT
strip between the infant's central chest area (in skin
© Oxford University Press 1998
15
N. MANANDHAR ET AL.
contact) and the mattress. A mercury-in-glass thermometer was placed in the axilla. Temperature readings
were taken contemporaneously after allowing lOmin for
equilibration.
LCT
In a liquid crystal, selective light scattering occurs at a
specific wavelength. The wavelength of maximum
scattering changes inversely with temperature. Thus, an
appropriately calibrated strip of liquid crystals appears to
'light up' in different places according to its temperature.
An experimental 'colour contact thermometer' utilizing
this principle has been manufactured by TALC (Teaching Aids at Low Cost, PO Box 49, St Albans, Herts, AL 1
5TX, United Kingdom, Fax: 441727 846852). This
reinforced plastic strip allows the visual recognition of
temperatures in 1°C gradations between 30° and 41°C.
Three separate bands, i.e. cold (35°C and below), normal
(36-37°C), and hot (38°C and above) are clearly
distinguished. With a foam backing it is robust and
easily cleaned before re-use.
Data analysis
Clinical details for each infant were recorded on a
standard form and data entered into a Macintosh
computer and analysed using Statview 4.1. As our gold
standard of neonatal hypothermia we used an axillary
temperature below 36°C measured by a mercury-in-glass
thermometer. The sensitivity, specificity, and likelihood
ratio for the detection of neonatal hypothermia was
calculated for the LCT method in comparison with the
gold standard. 95 per cent confidence intervals were
derived from the binomial distribution.
The likelihood ratio for a test result compares the
likelihood of that result in patients with disease (in this
paper disease is hypothermia) to the likelihood of that
result in patients without disease. Using estimates of the
pretest probability of neonatal hypothermia based on the
measured prevalence using mercury thermometry in this
population of first-day newborns, the likelihood ratio
was applied to a nomogram for Bayes theorem13'14 to
estimate the post-test probability of detecting hypothermia. This gives a measure of the usefulness of the
test for health workers.
Results
Clinical details
Table I describes the clinical details of the study infants.
Our study newborns had a similar distribution of
birthweight to that reported for a larger study of newborn
anthropometry in this setting.15
The mean SD, and range of newborn temperature in 61
infants measured by axillary mercury thermometry was
36.0, 0.68, 35.0-37.9°C; and by LCT was 35.8, 0.75,
35-37°C. Due to the unavailability of low reading
mercury thermometers this excludes a subgroup (n = 15)
whose temperature was below the range of a standard
16
mercury thermometer. The LCT values for this sub-group
were 34, 0.66, 32-55°C.
Comparison of LCT with mercury thermometry
The sensitivity, specificity and positive and negative
predictive values for the detection of neonatal hypothermia by LCT are shown in Table 2. The mercury
thermometer readings on the eight false negatives all lay
in the mild range of hypothermia (35.5-35.9°C) There
was a 63 per cent point prevalence of hypothermia
among this population of first day newborns using the
axillary mercury thermometry data. We took this value
to be the pretest probability of hypothermia among
newborns in this setting during the winter. Using the
likelihood ratio for the LCT method, applied to a
nomogram for Bayes theorem, the estimated post-test
probability of detecting hypothermia rose to 97 per
cent.14 If a much lower pre-test clinical probability of
hypothermia is assumed, e.g. 10 per cent, which might be
more appropriate in summer or at sea-level, the post-test
probability still exceeds 70 per cent.
Discussion
Coir study has shown that liquid crystal thermometry is a
valid method for identifying core hypothermia in
newborns. From a clinical standpoint, the use of LCT
to monitor newboms reliably identifies from 70 to 97 per
cent of hypothermic newborns depending upon the pretest probability. This suggests that LCT could be a useful
tool for use by paediatricians, midwives and other health
workers in the developing world.
The production cost of an LCT is comparable to that
of a mercury thermometer. At present LCTs are made for
TALC as a prototype, at a production cost of less than £ 1,
and are not commercially available. The LCT strip has
two important advantages compared to a mercury
thermometer it is not fragile and the colour coding is
easily understood by illiterate health workers or mothers
who may have difficulty with a graded scale. One caveat
is that the relative advantage of LCTs compared with
temperature assessment by touch alone has not yet been
assessed. Singh and colleagues in Delhi showed that
paediatricians using touch alone could identify hypothermia in newborns with a high level of accuracy which
correlated with the experience of the observer.16 It would
be interesting to evaluate the accuracy of 'touch
TABLE 1
Clinical details of the infants
76
Number of subjects
37:39
Gender male: female
2.79(2.00-3.70)0.41
Birth weight (kg)
mean (range) SD
Age at temperature assessment (h) 7(1-23)4.9
mean (range) SD
Ambient temperature (°C)
21.0 (14.0-28.2) 3.7
mean (range) SD
Journal of Tropical Pediatrics
Vol.44
February 1998
N. MANANDHAR ET AL.
TABLE 2
Detection of neonatal hypothermia (< 3(PC) by LCT compared with mercury
thermometry as a gold standard
Hypothermic (Hg)
Normothermic (Hg)
Totals
Hypothermic (LCT)
Normothermic (LCT)
40
8
1
27
41
35
Totals
48
28
76
% (95% CI)
Sensitivity
Specificity
Positive
predictive value
Negative
predictive value
Likelihood ratio
83 (70-92)
% (82-100)
98(86-100)
77 (59-89)
23.1
assessment' by midwives, nurses, and mothers, compared with using LCT strips. Pilot studies are also
needed to evaluate the use of LCTs by traditional
birth attendants, and whether addition of LCTs to
delivery kits should be considered in Safer Motherhood
programmes.
References
Ji XC, Zhu CY, Pang RY. Epidemiological study on
hypothermia in newboms. Chin Med J 1993; 106: 428-32.
Bnend A, de-Schampheleire I. Neonatal hypothermia in
West Africa (letter). Lancet 1981; 1: 846-7.
Karan S, Rao MN, Urmila S, Rajaji S. The incidence,
clinical profile, morbidity and mortality of hypothermia in
the newborn. Ind Pediat 1975; 12: 1205-10.
4. Ellis M, Manandar N, Shakya U, Manandar DS, Fawdry A,
Costello AM de L. Postnatal hypothermia and cold stress
among newborn infants in Nepal monitored by continuous
ambulatory recording. Arch Dis Child 1996; 75: F42-5.
5. Johanson RB, Spencer SA, Rolfe P, Jones P, Malla DS.
Effect of post-delivery care on neonatal body temperature.
Acta Paediat 1992; 81: 859-63.
6. Tafari N. Hypothermia in the tropics: epidemiologic
aspects. In: Sterky G, Tunell R, Tafari N (ed.) Judging
the appropriateness of technology. Stockholm: SAREC,
1985: 45-50. vol. R2).
Frank JD, Brown S. Thermometers and rectal perforations
in the neonates. Arch Dis Child 1978; 53: 284-5.
Journal of Tropical Pediatrics
Vol.44
February 1998
8. Im SW. Rectal thermometer mediated cross-infection with
Salmonella Wandsworth in a paediatric ward. J Hosp Infect
1981; 2: 171-4.
9. Mayfield SR, Bhatia J, Nakamura KT, Rios GR, Bell EF.
Temperature measurement in term and preterm infants.
J Pediat 1984; 104:271-5.
10. Schiffman RF. Temperature monitoring in the neonate. A
comparison of axillary and rectal temperatures. Nurs Res
1982; 31: 274.
11. WHO. Thermal Control of the Newborn: a practical guide.
In: Maternal Health and Safe Motherhood Programme,
Division of Family Health, 1993 WHO. pp. 3-17. (WHO/
FHE/MSM/93-2).
12. Valadez JJ, Elmore-Meegan M, Morley D. Comparing
liquid crystal thermometer readings and mercury thermometer readings of infants and children in a traditional
African setting: implications for community-based health.
Trop Geogr Med 1995; 47: 130-3.
13. Fagan TJ. Nomogram for Bayes theorem. N Engl J Med
1975; 293: 257.
14. Jaeschke R, Guyatt G, Sackett D. Users' guides to the
medical literature. III. How to use an article about a
diagnostic test JAMA 1994; 271: 703-7.
15. Manandhar DS, Rajbhandari S, Costello AM de L.
Anthropometry of the newborn and postnatal mother in
Nepal. J Nep Med Ass 1997; 35: 150-157.
16. Singh M, Rao G, Malhotra AK, Deorari AK. Assessment of newborn baby's temperature by human touch: a
potentially useful primary care strategy. Ind Pediat 1992;
29: 449-52.
17