Effect of nail polish on oxygen saturation determined by pulse oximetry in

Effect of nail polish on oxygen saturation determined by pulse oximetry in
Resuscitation (2007) 72, 82—91




CLINICAL PAPER


Effect of nail polish on oxygen saturation
determined by pulse oximetry in
critically ill patients夽
Jochen Hinkelbein ∗, Harald V. Genzwuerker, Reiner Sogl, Fritz Fiedler

University Clinic for Anesthesiology and Intensive Care Medicine, Faculty of Clinical Medicine Mannheim,
Ruprecht-Karls-University Heidelberg, University Hospital Mannheim, Theodor-Kutzer-Ufer 1-3,
D-68167 Mannheim, Germany

Received 16 February 2006 ; received in revised form 30 May 2006; accepted 13 June 2006


  KEYWORDS                            Summary
  Nail polish;                        Introduction: Nail polish of different colours may alter accuracy and precision of
  Pulse oximetry;                     pulse oximetry as previous data in healthy volunteers suggest. This trial evaluates
  Monitoring;                         the oxygen saturation determined by pulse oximetry and haematoximetry with nail
  Hematoximetry;                      polish of nine different colours applied.
  Bias;                               Material and methods: Fifty critically ill and mechanically ventilated patients in an
                                      ICU were investigated in a prospective clinical—experimental trial. On nine finger
  Functional oxygen
                                      nails polish of different colours was applied in a predetermined consecutive order.
  saturation;
                                      Functional oxygen saturation was determined by pulse oximetry (SpO2 ) on each finger
  Intensive Care Medicine;
                                      for each colour with the finger sensor probe both in the normal position and at a
  SaO2 ;                              90◦ rotation. Simultaneously oxygen saturation was determined by haematoximetry
  SpO2                                (SaO2 ). Accuracy (bias, S = SaO2 −SpO2 ) and precision (standard deviation, S.D.) of
                                      pulse oximetry were analyzed with the t-test. A value of P < 0.05 was considered
                                      significant.
                                      Results: While black (S = +1.6 ± 3.0%), purple (S = +1.2 ± 2.6%) and dark blue nail
                                      polish (S = +1.1 ± 3.5%, each N = 50) had the greatest effect (P < 0.05), all other
                                      colours, including colourless nail polish, had a smaller effect (mean bias +0.2 to
                                      +0.9%). A rotation of 90◦ reduced the bias from +2.8 to +1.3% (N = 10, n.s.).
                                      Conclusion: Nail polish does not alter pulse oximetry readings in mechanically ven-
                                      tilated patients to a clinically relevant extent. The mean error of measurement for
                                      all colours was within the manufacturers’ specified range of ±2%. A 90◦ rotation of
                                      the sensor probe does not eliminate errors in measurement. To remove nail polish
                                      might be helpful to decrease the error of measurement in some cases.
                                      © 2006 Elsevier Ireland Ltd. All rights reserved.



 夽 A Spanish translated version of the summary of this article appears as Appendix in the final online version at

10.1016/j.resuscitation.2006.06.024.
 ∗ Corresponding author. Tel.: +49 621 383 2415; fax: +49 621 383 732740.

   E-mail address: jochen.hinkelbein@anaes.ma.uni-heidelberg.de (J. Hinkelbein).
0300-9572/$ — see front matter © 2006 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.resuscitation.2006.06.024
Effect of nail polish on oxygen saturation determined by pulse oximetry in critically ill patients               83

Introduction                                                patients. Only a few studies, all performed with
                                                            a small number of volunteers, have substantiated
Since improvements to the pulse oximeter in the             these results. Studies involving mechanically venti-
early 1970s by Takuo Aoyagi from the Nihon Kohden           lated and critically ill patients, have not been pub-
Corporation, pulse oximetry has become a stan-              lished to date. Furthermore, newer and improved
dard monitoring technique during anaesthesia in             pulse oximeters and sensors and new nail polish
the operating room (OR), treatment in the ICU               products require further evaluation and an accu-
and emergency medical resuscitation of critically           racy validation of the bias induced by nail polish.
ill patients.1,6,9,21,28,33 Pulse oximeters determine          A prospective experimental clinical trial in
the percentage of oxygenated haemoglobin within             mechanically ventilated critically ill patients in an
pulsatile arterial blood, which is known as func-           intensive care unit was performed using haema-
tional (or partial) oxygen saturation. By means of          toximetry and pulse oximetry to evaluate the mea-
haematoximetry — widely considered a reference —            surements using the two techniques with different
the functional oxygen saturation may also be deter-         nail polish colours.
mined.
    During patient monitoring using pulse oximeters
the early detection of untoward events is most              Material and methods
important, as it may contribute to the prevention
of hypoxic insults,6 and therefore improve patient          Patients
safety.9,21 Since the deleterious effects of peri-
operative hypoxaemia has been recognised, the use           This study was approved by the local ethics com-
of pulse oximetry has become standard in operating          mittee for human research and was performed in
rooms.33                                                    accordance with the ethical standards laid down
    It must be noted however, that circumstances            in the 1964 Declaration of Helsinki. Relatives were
exist in which the pulse oximeter may give inac-            informed about the procedure and asked for writ-
curate or misleading information about arterial             ten informed consent prior to inclusion in the
oxygen saturation. Until now, many factors have             study.
been described which can lead to erroneous results             Data were collected on all patients of the ICU
in measurement of oxygen saturation by pulse                meeting inclusion within 1 year. The inclusion cri-
oximetry. One of these factors is the use of nail           teria were an age of at least 18 years, white skin
polish.8,12,13,27,34 Blue, green, and black coloured        colour (Caucasian race), and mechanical ventila-
nail polish are considered especially to result in          tion for at least 24 h via a tracheal tube. All patients
inaccurate measurements.                                    had been admitted to the ICU at least 24 h before
    Today the effect of nail polish on pulse oximetry       inclusion. Five fingers on each upper extremity
is discussed extensively but has not been inves-            without nail polish applied prior to the trial were
tigated in critically ill or mechanically ventilated        essential to participate. Carboxyhaemoglobin and




       Figure 1 Absorption spectra of all colours used: (1) black, (2) dark green, (3) purple, (4) dark blue.
84                                                                                                 J. Hinkelbein et al.




Figure 2 Absorption spectra of all colours used: (5) red, (6) light blue, (7) colourless, (8) light green, (9) yellow.

methaemoglobin values each had to be less than                 and module SIREM, Siemens Medical Electronics,
4.0% when determined by haematoximetry.                        Danvers/USA) connected to a re-usable finger sen-
   If the plethysmographic waveform was of poor                sor probe NELLCORTM DS-100A Durasensor® (Nell-
quality (by visual analysis of a flat or irregu-                cor Puritan Bennett Inc., Pleasanton, USA). This
lar signal waveform or difference to ECG heart                 sensor probe contains two different light emit-
rate of more than 3 beats per minute), the                     ting diodes (LED), emitting light at wavelengths
patient was excluded. If there was a nail myco-                of 660 and 920 nm, and a photo detector diode
sis, severe anaemia (Hb < 6.0 g/dl), severe hypoten-           (PDD) that measures light passing through the
sion (systolic blood pressure 40 mg/dl),                 To facilitate assignment of colours, the fingers
or hypothermia (temperature < 34 ◦ C), the patient             were numbered consecutively from left to right
was also excluded because interference had been                (thumb of left hand = no. 1, second finger of the left
reported previously for these factors when using               hand = no. 2, middle finger of the left hand = no. 3,
pulse oximetry.18,23,24,28,31,32                               and thumb of right hand = no. 6, etc.). On 9 of 10 fin-
                                                               gers, nail polish of a different colour was applied in
Nail polish                                                    two coats by the same investigator. Order of appli-
                                                               cation was standardised to ensure that among the
Due to the great number of nail polish brands                  patients, each finger was studied at least once with
and types available, it was impractical to test all            each colour:
variations in colour and pigment composition. We                  Subject #1: finger 1 = colour 1; finger 2 = colour
selected nine different colours of nail polish over            2; finger 3 = colour 3; and so on. No nail polish was
the full colour spectrum: black, dark blue, light              applied on finger 10. For subject #2, the colours
blue, dark green, light green, yellow, red, pur-               and fingers were changed: finger 1 = colour 2; finger
ple, and colourless (Table 1) available internation-           2 = colour 3; finger 3 = colour 4; and so on. No nail
ally. The spectral analysis for each colour (maxi-             polish was applied on finger 9, and finger 10 = colour
mum peak and minimum absorption wavelength)                    1. This rotation sequence was applied for all further
was performed three times with a spectrophotome-               patients.
ter (LS 500, DRLANGE, Germany) between 500 and                    Each coat was allowed to dry completely
950 nm (Figures 1 and 2) analysing standardised two            for 2 min between the applications (completely
layers of nail polish. None of the suppliers of nail           dry and not sticky; tested by the investigator).
polish were involved in this study.                            After the second coat was completely dried,
                                                               the measurements were taken. The remaining
Measurement                                                    finger without nail polish served as a control
                                                               for the SpO2 of the fingers with nail polish
All readings with pulse oximetry were obtained                 applied and additionally for the SaO2 obtained by
with a SIEMENS pulse oximeter (monitor SC1281                  haematoximetry.
Effect of nail polish on oxygen saturation determined by pulse oximetry in critically ill patients
Table 1        Characteristics of different colours used
Colour                    Black              Dark blue          Light blue         Dark green        Light green        Yellow         Red              Purple             Colourless        None
Name                      Diamant            Diamant            FABIANI no. 2      Super Quick Dry   FABIANI no. 3      KRON 79        MISSLYN no. 34   FABIANI no. 5      CHICOGO 581       n.a.
                          Sensation 975      Sensation 550                         no. 85                                                                                  Clear
Manufacturer              COTY ASTOR,        COTY ASTOR,        Dobner Kosmetik,   Lady Manhattan    Dobner Kosmetik,   INTERCO,       INTERCO,         Dobner Kosmetik,   Rimmel, London,   n.a.
                          Barcelona, Spain   Barcelona, Spain   Albaching,         Cosmetics GmbH,   Albaching,         Wiesbaden,     Wiesbaden,       Albaching,         UK
                                                                Germany            Stuttgart,        Germany            Germany        Germany          Germany
                                                                                   Germany
Wavelength (peak max,     830                690                710                720               500                640            550              530                890               n.a.
  nm)
Wavelength min (nm)       520                500                950                510               950                500            950              950                500               n.a.
SpO2 ± S.D. (%)           96.2 ± 3.2         96.6 ± 3.6         96.8 ± 2.7         96.9 ± 2.5        97.2 ± 2.7         97.0 ± 2.4     96.9 ± 2.4       96.6 ± 2.9         97.1 ± 2.5        97.5 ± 2.2
Bias ± S.D. (%)           1.6 ± 3.0          1.1 ± 3.5          0.9 ± 2.2          0.9 ± 2.3         0.6 ± 2.3          0.8 ± 2.1      0.8 ± 2.0        1.2 ± 2.6          0.7 ± 2.2         0.2 ± 1.5
Bias maximum              10.6%              9.1%               8.5%               9.1%              9.5%               6.5%           5.4%             12.5%              7.5%              4.1%
Bias minimum              −4.1%              −5.7%              −2.9%              −4.4%             −5.1%              −3.1%          −2.9%            −3.1%              −5.1%             −2.9%
Bias in the defined        26 (52%)           23 (46%)           31 (62%)           34 (68%)          39 (78%)           36 (72%)       36 (72%)         34 (68%)           35 (70%)          42 (84%)
  interval, S < ±2%
Bias, S > +2% (n)/%      19 (38%)           18 (36%)           15 (30%)           13 (26%)          9 (18%)            12 (24%)       12 (24%)         14 (28%)           12 (24%)          6 (12%)
Bias, S < −2% (n)/%      2 (4%)             7 (14%)            3 (6%)             2 (4%)            2 (4%)             1 (2%)         2 (4%)           1 (2%)             3 (6%)            2 (4%)
Limits of agreement:      1.6 ± 6.0          1.1 ± 7.2          0.9 ± 4.4          0.9 ± 4.6         0.6 ± 4.6          0.8 ± 4.2      0.8 ± 4.0        1.2 ± 5.2          0.7 ± 4.4         0.2 ± 3.0
  S ± 2S.D. (%)
Limits of agreement (%)   −4.4 to +7.6       −6.1 to +8.3       −3.5 to +5.3       −3.7 to +5.5      −4.0 to +5.2       −3.4 to +5.0   −3.2 to +4.8     −4.0 to +6.4       −3.7 to +5.1      −2.8 to 3.2
Limits of agreement:      12.0               14.4               8.8                9.2               9.2                8.4            8.0              10.4               8.8               6.0
  dimension (%)
Total number (n)          50 (100%)          50 (100%)          50 (100%)          50 (100%)         50 (100%)          50 (100%)      50 (100%)        50 (100%)          50 (100%)         50 (100%)
Significance of SpO2 to    P < 0.002          P < 0.04           P < 0.008          P < 0.02          n.s.               P < 0.02       P < 0.006        P < 0.004          P < 0.04          n.s.
  SaO2




                                                                                                                                                                                                           85
86                                                                                         J. Hinkelbein et al.

   The finger sensor probe was applied directly to         of measurement error were calculated.3,29 As an
the center of the fingernail on each forefinger to          indicator for the spread of readings, the difference
ensure that light passed the finger from dorsal to         between the minimum bias and the maximum bias
volar side (standard position, top-to-bottom posi-        was calculated for each colour. A value of P < 0.05
tion). When stable pulse oximeter readings of oxy-        was considered significant.
gen saturation were obtained for at least 20 s, the          Statistical analysis was performed with
reading was recorded.                                     STATISTIKA® (V 6.0, StatSoft Inc., USA). The
   In 10 randomly selected patients, the probe was        T-test for paired values was used to analyze the
additionally rotated by 90◦ , so that the optical path    significance of variables. Values are expressed as
cut laterally through the finger and not through the       means ± standard deviation (range).
nail bed (side-to-side position). When a stable mea-
surement of oxygen saturation was reached for at
least 20 s another reading in the lateral position was    Results
recorded.
   At the same time as the pulse oximetry mea-            Patient characteristics
surement, an arterial blood sample for analysis was
drawn by a second person. The time-point was stan-        A total of 50 critically ill patients with ventilation
dardised and scheduled between the pulse oxime-           support (19 female and 31 male) in our ICU were
try measurement at finger no. 5 and finger no. 6.           included in this clinical study, representing 500 data
The blood samples were obtained in all patients           points (i.e. 10 fingers and readings each) between
via an arterial catheter already in place as part         August 1st and December 31st , 2003. A total of 212
of the routine clinical management in the right           patients had to be excluded because they were
or left radial artery and functional arterial oxy-        not mechanically ventilated (N = 184), presented
gen saturation (SaO2 ) was determined optically by a      no sufficient plethysmographic pulse wave signal
haematoximeter (Radiometer Copenhagen ABL Sys-            (N = 16), relatives declined to participate (N = 6),
tem 625, Radiometer, Brønshøj, Denmark). For fur-         were younger than 18 years (N = 4) or were not
ther analysis, the functional oxygen saturation of        of Caucasian race (N = 2). No patient had to be
both techniques was compared.                             excluded due to dyshaemoglobinemia, nail myco-
                                                          sis or amputation of fingers.
Statistical data analysis                                    All participating patients suffered from surgical
                                                          diseases and surgical interventions as the reasons
The functional oxygen saturation (SaO2 ) is defined        for admission to the ICU (Table 2). The mean
as the concentration of oxyhaemoglobin divided by         fraction of inspired oxygen (FiO2 ) was 49 ± 15%
the total present haemoglobin able to carry oxygen:       (range, 21—100%) resulting in a mean arterial
SaO2 = oxyhaemoglobin [O2 Hb]/(oxyhaemoglobin             partial oxygen pressure of 104.6 ± 29.2 mmHg
[O2 Hb] plus desoxyhaemoglobin [dHb]).22 SaO2 is          (range, 73.1—213.1 mmHg) and a measured
usually used as a reference for functional oxygen         mean oxygen saturation SaO2 of 97.8 ± 1.3%
saturation determined by pulse oximetry (SpO2 ).          (range, 94.3—100.0%). The mean temperature was
   Statistical analysis of the data focused on the dif-   37.1 ± 0.4 ◦ C (range, 36.4—38.1 ◦ C, N = 50), the
ference in oxygen saturation between the fingers to        signal quality of pulse oximetry readings was not
which nail polish had been applied, and the corre-        impaired by severe hypothermia.
sponding finger without nail polish in the subject,
and the oxygen saturation determined by haema-            Analysis of accuracy
toximetry.
   To assess the congruence of measurement                All colours of nail polish showed a different impair-
between the two techniques, different analytical          ment of the readings of pulse oximetry. Mean SaO2
methods were used: the method described by Bland          (97.8 ± 1.3%, range 94.3—100.0%, N = 50) correlated
and Altman3 was used to calculate the mean differ-        well with the mean SpO2 of the unpainted finger
ence between SaO2 and SpO2 and hence measure-             (S = +0.2 ± 1.5% [range, −2.9 to +4.1%], N = 50,
ment accuracy (accuracy, bias, S = SaO2 − SpO2 ).        n.s.).
To assess the precision of measurements, stan-               While black (S = +1.6 ± 3.0% [range, −4.1
dard deviation of the differences (S.D., variability)     to +10.6%], P < 0.002), purple (S = +1.2 ± 2.6%
between SpO2 and SaO2 , the number of outliers            [range, −3.1 to +12.5%], P < 0.004) and dark blue
(bias-values above and below the manufacturers’           nail polish (S = +1.1 ± 3.5% [range, −5.7 to +9.1%],
given interval of ±2%), and the limits of agree-          P < 0.04) revealed the greatest effect, all other
ment (bias ± 2S.D.) in all used colours as an index       colours reached a mean bias below 1% (Table 1).
Effect of nail polish on oxygen saturation determined by pulse oximetry in critically ill patients                         87

 Table 2    Demographical and epidemiological data of all participating patients
                                                                                                Number (N) (%)
 Sex
   Male (N)                                                                                     31 (62%)
   Female (N)                                                                                   19 (38%)
 Diagnosis
   Subarachnoid hemorrhage (N)                                                                  12 (24%)
   Trauma (N)                                                                                   7 (14%)
   Carcinoma (N)                                                                                5 (10%)
   ARDS (N)                                                                                     4 (8%)
   Shock (N)                                                                                    4 (8%)
   Sepsis (N)                                                                                   2 (4%)
   Other (N)                                                                                    16 (32%)
 Ventilation
   Intubation/tracheostomy and mechanical ventilation (N)                                       50 (100%)
 Demographic
   Age (years)                                                                                  59 ± 14.2 (20—82)
 Ventilation
   FiO2 (%)                                                                                     49.4 ± 15.3 (21.0—100.0)
 Circulation
   Systolic blood pressure (mmHg)                                                               138 ± 27 (86—199)
   Temperature (◦ C)                                                                            37.1 ± 0.4 (36.4—38.1)
 Haematoximetry and arterial blood gases
   Oxygen saturation (SaO2 ) (%)                                                                97.8 ± 1.3 (94.3—100.0)
   paO2 (mmHg)                                                                                  104.6 ± 29.2 (73.1—213.1)
   paCO2 (mmHg)                                                                                 43.9 ± 6.9 (29.9—60.0)
   pH                                                                                           7.438 ± 0.067 (7.259—7.624)
   COHb (%)                                                                                     1.6 ± 0.6 (0.4—3.1)
   MetHb (%)                                                                                    0.8 ± 0.3 (0.2—1.8)
   Haemoglobin (g/dl)                                                                           9.0 ± 1.5 (6.4—12.0)
   Plasma bilirubin (mg/dl)                                                                     1.96 ± 3.85 (0.17—19.88)
 Subgroups
   Patients with parenteral nutrition (lipids) (ml/h), N = 20                                   70 ± 18 (42—110)
   Patients with Noradrenaline (nor epinephrine) (␮g/kg/min), N = 17                            22.4 ± 29.3 (1.3—100.0)
 Data are given as number (N) and percentage (%) or as mean ± standard deviation (S.D.) and range.


   Light green nail polish (S = +0.6 ± 2.3% [range,                 There was a wide range for the spread of each
−5.1 to +9.5%], n.s.) had the smallest effect of                  of the colours’ bias values (minimum bias to maxi-
all colours used. With the exception of light green               mum bias of each reading) with a maximum of 15.6%
and the unpainted nail, a statistically significant                (range, −3.1 to +12.5%) with the colour purple. The
difference to the SaO2 was found for all colours                  minimum range of bias values (8.3%) was achieved
(Figure 3).                                                       with red nail polish (range, −2.9 to +5.4%).
   The mean bias was always inside the specified
limit of ±2% and was positive (S > 0%) for all mea-
surements implicating that pulse oximetry usually
                                                                  Analysis of outliers
underestimates the actual oxygen saturation deter-                Concerning the outliers (i.e. the number of bias val-
mined by haematoximetry (SaO2 > SpO2 ).                           ues over and below the specified manufacturers’
                                                                  range of ±2%20 ), the results of our study show that
Analysis of precision                                             SpO2 usually tends to underestimate SaO2 in criti-
                                                                  cally ill ventilated patients with nail polish applied.
The standard deviation of functional oxygen satu-                 Of all readings, a total of N = 130 readings (i.e.
ration in each colour ranged from 2.0% (red nail                  26 ± 7.7% [range, 12—38%] from N = 500) had a bias
polish) to 3.5% (dark blue nail polish). The standard             greater than the specified +2% with a SpO2 lower
deviation at the unpainted finger was lowest (1.5%).               than SaO2 . In contrast, only N = 25 readings (i.e.
88                                                                                         J. Hinkelbein et al.




Figure 3 Mean bias depending on the colour of nail polish. The whiskers show maximum and minimum bias, the thick
black horizontal line the mean bias. Top and bottom of the box describe the 25 and 75% quartiles.

5 ± 3.4% [range, 2—14%] from N = 500) had a bias          Discussion
below −2% (SpO2 higher than SaO2 ).
   With black nail polish, the bias was greater than      Takuo Aoyagi was the first engineer to develop an
+2% in N = 19 (38%) measurements while with dark          apparatus and sensor able to perform real-time
blue there were N = 18 (36%) measurements out-            pulse oximetry for routine clinical use.1,28 Today
side the manufacturers interval. The fewest out-          pulse oximetry has become a standard monitoring
liers >+2% were found (N = 6, 12%) if no nail polish      technique during anaesthesia, intensive care ther-
was applied.                                              apy and emergency medical treatment.
                                                             Many factors which can affect values of oxygen
Analysis of the limits of agreement                       saturation obtained by pulse oximetry have been
                                                          previously described: extreme anaemia,15,16,18,31
The dimension of the limits of agreement was cal-         pharmacotherapy         with     catecholamines,15,31
culated as the bias ± 2S.D. The limits of agreement       carboxyhaemoglobinaemia and elevated COHb-
range between 6.0% for the unpainted fingernail            levels,2,4,19,23 dried blood at sensor site,26
indicating a small variation of all readings, and         hyperbilirubinaemia,23,32        hypotension      and
14.4% for the dark blue colour nail polish, indicating    hypoperfusion,11,15,31 hypoxia,28,30 intravenous
a great variation of the differences.                     dyes,23 motion23 and methaemoglobinaemia and
                                                          elevated MetHb level.10,23 Additionally, the sensor
Analysis of reliability and variability after             site influences measurement due to variable kinet-
90◦ rotation                                              ics, although the effect is small and may be not
                                                          clinically relevant.7,14,25
In the 10 patients randomly selected for evalua-             Nail polish has been reported to interfere
tion of a 90◦ rotation of the finger sensor probe          with pulse oximetry.5,8,27,34 There is ongoing
(five male, five female, mean age 64 ± 12 [range,           controversy,17 especially since all studies per-
42—76] years) a mean SaO2 of 97.7 ± 1.1% [range,          formed to date only involved healthy volunteers but
95.6—99.3%] was measured. In 40 out of a total            did not include critically ill patients on ventilatory
of 100 values, the calculated bias was above +2%          support.
or below −2%. These data sets were chosen for                Theoretically, there should be no influence
further investigation. In these data pairs, the bias      of nail polish on pulse oximetry, because pulse
was calculated as S = +2.8 ± 3.7% [range, −2.8 to        oximetry uses the pulsatile light-absorption of
+9.5%] (N = 40), and after rotation of the finger sen-     haemoglobin to determine oxygen saturation. How-
sor probe by 90◦ as S = +1.3 ± 2.2% [−1.4 to +4.6%]      ever, our results show that there is an interference
(N = 40, P > 0.05).                                       that is most significant for black, purple and dark
Effect of nail polish on oxygen saturation determined by pulse oximetry in critically ill patients           89




          Figure 4 Overall relationship between SaO2 (x-axis) and SpO2 (y-axis) at the unpainted finger.




blue nail polish. All other colours, and colourless       sibly black and brown-red nail polishes interfere
nail polish, had a lesser effect on pulse oximetry        with pulse oximetry. These colours reduced the
readings.                                                 SpO2 reading up to 6% with an underestimation of
   In all colours tested, pulse oximetry (SpO2 )          SaO2 . This effect was clinically relevant and was
tended to underestimate the oxygen saturation             attributed to the fact that these colours had higher
determined by haematoximetry (SaO2 ). Readings            absorbance near 660 nm causing the oximeter to
obtained by pulse oximetry were generally lower           artificially detect a higher proportion of desoxy-
compared to SaO2 readings if nail polish is applied,      haemoglobin. The authors concluded that removal
while values correlated well when no nail polish is       of nail polish is mandatory before a pulse oximeter
applied (Figure 4). A possible explanation for the        sensor is attached to prevent erroneous oximeter
slight decrease in oxygen saturation readings with        readings.8
nail polish present could be a decrease of pulse             Brand et al.5 also tested healthy subjects. Ten
wave signal quality due to an increased absorption        different colours of nail polish with three coats
of light.                                                 were applied on the three inner fingers. The index
   Most manufacturers (e.g. SIEMENS for the SC1281        finger of the opposite side was used for control mea-
used in this study) specify that their pulse oxime-       surements. In contrast to Coté et al.,8 Brand et al.5
ter readings are reliable with a deviation of ±2% in      found impairment of not more than 1% in green and
the saturation range from 70 to 100%.20 Even with         blue nail polish.
black, purple and dark blue nail polish applied on           Important limitations of all the studies described
the fingernails of critically ill patients, the devia-     are the lack of patient data and the small number
tion or mean bias was within the specified range of        of subjects studied. There is no study published to
±2%. Average measurements were always within a            date investigating the influence of nail polish on
clinically acceptable range.                              pulse oximetry readings in critically ill patients. In
   First reports on the influence of nail polish on        addition, no study is available which includes crit-
pulse oximetry can be found in letters from the mid       ically ill patients on ventilation support and eval-
to late 1980s17,27 and the results described were         uates the error in measurement with nail polish in
controversial. Kataria and Lapkins found no signifi-       this patient group.
cant error in the measurement in 15 volunteers with          In mechanically ventilated and critically ill
nail polish applied, and therefore stated that nail       patients with multiple factors influencing pulse
polish does not affect pulse oximeter saturation.17       oximetry, accurate readings are essential. In this
Rubin et al. described effects of nail polish of          population a lower accuracy and lower precision of
31 different colours on oxygen saturation deter-          the measurement (i.e. due to application of cate-
mined by pulse oximetry. However, one female              cholamines, haemodynamic changes, etc.) is possi-
volunteer was the sole subject of this study.27           ble. Since the available data obtained in healthy
Both studies did not quantify the degree of the           volunteers may not be applicable to critically ill
impairment.                                               and mechanically ventilated patients, we evaluated
   Coté et al.8 investigated 14 male and female          the accuracy and precision of pulse oximetry in this
volunteers and reported that blue, green and pos-         patient population.
90                                                                                            J. Hinkelbein et al.

   The comparison of measurements of pulse              Conclusion
oximetry with the oxygen saturation determined by
haematoximetry in fingers with nail polish has also      The nine colours of nail polish tested did not inter-
not been described so far.                              fere with pulse oximetry to a clinically relevant
   Our trial identified black, purple, and dark blue     extent. The mean error of measurement for all
as the most influencial colours. Coté et al.8 found     colours was within the manufacturers’ specified
black and blue to have a significant influence as         range of ±2%. The actual oxygen saturation tends
well. Blue was also identified by Brand et al.5 In       to be underestimated if nail polish is applied com-
addition, both authors reported green to lead to a      pared to SaO2 measurements.
significant bias. In this trial we found dark green         Clinically significant deviations in pulse oximeter
(not light green) to influence the measurement sig-      readings rarely seem to be caused by nail polish. To
nificantly, but with a smaller bias than black, purple   remove nail polish might be helpful to decrease the
and dark blue.                                          error of measurement in some cases.
   While Brand et al.5 reported the bias to be below
1%, Coté et al.8 found the bias to be up to 6%.
   Consequently, our results from critically ill        Acknowledgements
patients are largely in accordance with Coté et
al.8 However, in contrast to Coté et al.8 we cannot    No financial or material support was received for
recommend that nail polish generally is removed,        this study from the companies mentioned. We wish
because the mean bias is below 2% compared to           to thank the patients and staff of our ICU as well as
haematoximetry, although there were some severe         the members of the Institute of Clinical Chemistry
outliers with a great bias.                             at the University Hospital Mannheim who performed
                                                        the photo spectral analysis of the nail polishes, for
Elimination of faulty results                           their support.
                                                           Preliminary results were presented at The Sec-
To eliminate faulty pulse oximetry results in sub-      ond Mediterranean Emergency Medicine Congress
jects with nail polish applied, White et al.34 and      2003 at Sitges/Spain13 and at the Annual Meeting
Brand et al.5 recommended rotating the sensor           of the European Society of Anaesthesiologists 2004
probe by 90◦ . We found that a rotation of the fin-      in Lisboa,12 as both an abstract and oral presenta-
ger sensor probe by 90◦ from the standard posi-         tion.
tion can reduce faulty readings, but the changes
are without clinical and statistical significance. In
contrast to the results of other authors,5,34 our       References
results revealed no significant change in reliability,
variability and accuracy of pulse oximetry readings     1. Aoyagi T, Kishi M, Yamaguchi K, Watanabe S. Improvement
after 90◦ rotation; the underlying reasons remained        of an ear piece oximeter. In: Abstracts of the 13th annual
unclear.                                                   meeting of the Japanese Society for Medical Electronics and
                                                           Biological Engineering. 1974. p. 90—1.
                                                        2. Barker SJ, Tremper KK. The effect of carbon monoxide
Limitations                                                inhalation on pulse oximetry and transcutaneous PO2 . Anes-
                                                           thesiology 1987;66:677—9.
We tested nine different colours of nail polish in      3. Bland JM, Altman DG. Statistical methods for assessing
50 critically ill ventilated patients. Different man-      agreement between two methods of clinical measurement.
ufacturers are assumed to use different chemical           Lancet 1986;I:307—9.
                                                        4. Bozemann WP, Myers RAM, Barish RA. Confirmation of the
components. Although none of the nail polishes             pulse oximetry gap in carbon monoxide poisoning. Ann Emerg
tested had a clinically significant effect on pulse         Med 1997;30:608—11.
oximetry readings, other nail polishes with differ-     5. Brand TM, Brand ME, Jay GD. Enamel nail polish does not
ent chromatogenes may influence the accuracy of             interfere with pulse oximetry among normoxic volunteers. J
the pulse oximeter to a greater extent. If nail pol-       Clin Monit Comput 2002;17:93—6.
                                                        6. Cheng EY, Hopwood MB, Kay J. Forehead pulse oximetry
ish is not applied in the standardised two layers as       compared with finger pulse oximetry and arterial blood gas
in our study, the influence on the accuracy of pulse        measurement. J Clin Monit 1988;4:223—6.
oximetry may also vary.                                 7. Clayton DG, Webb RK, Ralston AC, Duthie D, Runciman WB.
   Unfortunately, we were only able to perform the         A comparison of the performance of 20 pulse oximeters
study in Caucasian patients with a SaO2 between            under conditions of poor perfusion. Anaesthesia 1991;46:
                                                           3—10.
94.3 and 100%. Our findings may therefore be not         8. Coté CJ, Goldstein EA, Fuchsman WH, Hoaglin DC. The
applicable to patients with a SaO2 outside this range      effect of nail polish on pulse oximetry. Anesth Analg
or non-Caucasian patients.                                 1988;67:683—6.
Effect of nail polish on oxygen saturation determined by pulse oximetry in critically ill patients                                   91

 9. Eichhorn JH. Prevention of intraoperative anesthesia acci-              randomized, controlled trials. Anesth Analg 2003;96:
    dents and related severe injury through safety monitoring.              426—31.
    Anesthesiology 1989;70:572—7.                                     22.   Pologe JA. Functional saturation versus fractional sat-
10. Eisenkraft JB. Pulse oximeter desaturation due to methe-                uration: what does pulse oximetry read? J Clin Monit
    moglobinemia. Anesthesiology 1988;68:279—82.                            1989;5:298—9.
11. Hinkelbein J, Genzwuerker HV, Fiedler F. Detection of a sys-      23.   Ralston AC, Webb RK, Runciman WB. Potential errors in pulse
    tolic pressure threshold for reliable readings in pulse oxime-          oximetry III: effects of interference, dyes, dyshaemoglobins
    try. Resuscitation 2005;64:315—9.                                       and other pigments. Anaesthesia 1991;46:291—5.
12. Hinkelbein J, Genzwuerker HV, Fiedler F. How nail polish          24.   Ramsing T, Rosenberg J. Pulse oximetry in severe anaemia.
    alters oxygen saturation determined by pulse oximetry in                Intens Care Med 1992;18:125—6.
    the ICU setting. Eur J Anaesthesiol 2004;21:42 (abstract).        25.   Reynolds LM, Nicolson SC, Steven JM, Escobar A, McGonigle
13. Hinkelbein J, Genzwuerker HV, Fiedler F. Nail polish: bias              ME, Jobes DR. Influence of sensor site location on pulse
    evaluation and accuracy validation for the measurement of               oximetry kinetics in children. Anesth Analg 1993;76:751—4.
    oxygen saturation determined by pulse oximetry. J Emerg           26.   Rosewarne FA, Reynolds KJ. Dried blood does not affect pulse
    Med 2004;26:377 (abstract).                                             oximetry. Anaesthesia 1991;46:886—7.
14. Hinkelbein J, Hose D, Fiedler F. Comparison of three differ-      27.   Rubin AS. Nail polish color can affect pulse oximeter satura-
    ent sensor sites for pulse oximetry in critically ill patients.         tion. Anesthesiology 1988;68:825.
    Int J Intens Care 2005;12:159—63.                                 28.   Severinghaus JW, Astrup PB. History of blood gas analysis.
15. Hinkelbein J, Osika A, Fiedler F. Accuracy and precision of             Int Anesthesiol Clin 1987;25:1—224.
    pulse oximetry in ventilated patients of an intensive care        29.   Severinghaus JW, Kellerher JF. Recent developments in pulse
    unit. Eur J Anaesthesiol 2004;21:179 (abstract).                        oximetry. Anesthesiology 1992;76:1018—38.
16. Jay GD, Huges L, Renzi FP. Pulse oximetry is accurate in acute    30.   Severinghaus JW, Naifeh KH, Hoh SO. Errors in 14 pulse
    anemia from hemorrhage. Ann Emerg Med 1994;24:32—5.                     oximeters during profound hypoxia. J Clin Monit 1989;5:
17. Kataria BK, Lapkins R. Nail polish does not affect pulse                72—81.
    oximeter saturation. Anaesth Analg 1986;65:824.                   31.   Severinghaus JW, Spellman MJ. Pulse oximeter failure
18. Lee S, Tremper KK, Barker SJ. Effects of anemia on pulse                thresholds in hypotension and vasoconstriction. Anesthesi-
    oximetry and continuous mixed venous haemoglobin satura-                ology 1990;73:532—73.
    tion monitoring in dogs. Anesthesiology 1991;75:118—22.           32.   Taylor MB, Whitwam JG. The current status of pulse oxime-
19. Lee WW, Mayberry K, Crapo R, Jensen RL. The accuracy of                 try. Anaesthesia 1986;41:943—9.
    pulse oximetry in the emergency department. Am J Emerg            33.   van de Louw A, Cracco C, Cerf C, Harf A, Duvaldestin P,
    Med 2000;18:427—31.                                                     Lemaire F, et al. Accuracy of pulse oximetry in the inten-
20. Moyle JTB. Pulse oximetry. 2nd ed. London: BMJ Books; 2002.             sive care unit. Intens Care Med 2001;27:1606—13.
21. Pedersen T, Moller AM, Pedersen BD. Pulse oxime-                  34.   White PF, Boyle WA. Nail polish and oximetry. Anesth Analg
    try for perioperative monitoring: systematic review of                  1989;68:545—7.
You can also read
Next part ... Cancel