European Journal of Clinical Nutrition (2011) 65, 614–618
& 2011 Macmillan Publishers Limited All rights reserved 0954-3007/11
www.nature.com/ejcn
ORIGINAL ARTICLE
Diagnosing lactase deficiency in three breaths
M Oberacher1, D Pohl2, SR Vavricka1, M Fried2 and R Tutuian2,3
1
Department of Internal Medicine, Stadtspital Triemli, Zurich, Switzerland; 2Division of Gastroenterology and Hepatology,
Department of Internal Medicine, University Hospital Zurich, Zurich, Switzerland and 3University Clinics of Visceral Surgery
and Medicine, Inselspital Bern, Bern, Switzerland
Background: Lactose hydrogen breath tests (H2-BTs) are widely used to diagnose lactase deficiency, the most common cause of
lactose intolerance. The main time-consuming part of the test relates to the sampling frequency and number of breath samples.
Aim: Evaluate sensitivities and specificities of two- and three-sample breath tests compared with standard breath sampling every
15 min.
Methods: Lactose H2-BT with probes samples every 15 min served as gold standard. Sensitivity, specificity, positive and negative
predictive value of two-sample tests (0–60 min, 0–90 min or 0–120 min) and three-sample tests (0–60–90 min, 0–60–120 min or
0–90–120 min) were calculated.
Results: Among 1049 lactose H2-BT performed between July 1999 and December 2005, 337 (32%) had a positive result. Twosample tests had sensitivity and specificity of 52.5 and 100.0% (0–60 min), 81.9 and 99.7% (0–90 min), and 92.6 and 99.2%
(0–120 min), respectively. Three-sample tests had sensitivity and specificity of 83.4 and 99.7% (0–60–90 min), 95.0 and 99.2%
(0–60–120 min), and 95.0 and 98.9% (0–90–120 min), respectively.
Conclusion: A three-sample breath test (baseline, 60/90 min and 120 min) has excellent sensitivity and specificity for lactase
deficiency. Lactose H2-BT can be simplified but not shortened to o2 h.
European Journal of Clinical Nutrition (2011) 65, 614–618; doi:10.1038/ejcn.2010.287; published online 19 January 2011
Keywords: lactose intolerance; lactose hydrogen breath test; lactase deficiency; hypolactasia
Introduction
Lactose intolerance is a common condition affecting a large
proportion of the world’s population (Bayless et al., 1975).
Lactose, a disaccharide, is a carbohydrate present only in
mammalian milk accounting for around 7.2 g/l in human
milk and 4.7 g/l in cow’s milk (Solomons, 2002). The
prevalence of lactose intolerance differs in various parts of
the world ranging from 3 to 8% in the Scandinavian and
northwestern European population to 50–100% of the Asian
population (Anh et al., 1977; Densupsoontorn et al., 2004;
Farup et al., 2004). In Europe, the prevalence increases from
north to south and west to east reaching a peak of 70% in
southern Italy and Turkey (Gudmand-Hoyer, 1994).
The most common cause of lactose intolerance is lactase
deficiency, a condition of decreased production of lactase, the
enzyme in the small intestinal villi responsible for splitting
Correspondence: Dr R Tutuian, University Clinics of Visceral Surgery and
Medicine, Inselspital Bern, Freiburgerstrasse, Bern CH-3010, Switzerland.
E-mail: [email protected]
Received 30 August 2010; revised 15 November 2010; accepted 30
November 2010; published online 19 January 2011
lactose into the absorbable monosaccharides glucose and
galactose. In patients with lactose deficiency, the lactose
reaches the large intestine where it is metabolized by the
colonic flora. The high osmotic load caused by lactose in the
small intestine and its bacterial metabolites produced mainly
in the colon are considered to have an important role in the
pathogenesis of classic symptoms of lactose intolerance such
as diarrhea, bloating, nausea, borborygmi and abdominal pain.
There are different methods in diagnosing lactase deficiency, including small bowel biopsy, measurement of the
stool pH, oral administration of lactose followed by serial
measurements of blood glucose, genetic analysis (Enattah
et al., 2002; Pohl et al., 2010) or breath hydrogen concentration. The lactose hydrogen breath test (H2-BT) has been used
for more than 30 years, and is the preferred method to
diagnose lactase deficiency in clinical practice (Newcomer
et al., 1975). This test exploits the fact that, in contrast to
human metabolism, the colonic flora metabolizes lactose
into hydrogen (H2), methane and short chain fatty acids.
Hydrogen reaches the splanchnic venous circulation by
diffusion through the intestinal wall, is transported from
here through the portal system to the liver and the systemic
circulation and is eventually exhaled through the lungs.
Diagnosing lactase deficiency in three breaths
M Oberacher et al
615
Detecting an elevated H2 concentration in the exhaled air
has a good sensitivity (76–94%) and specificity (77–96%) to
diagnose lactase deficiency (Rosado and Solomons, 1983).
In its classical form, a lactose H2-BT requires collecting
breath samples at every 15–30 min over a 3- to 4-h time
period. In most laboratories this requires a dedicated
technician collecting and measuring breath samples over a
given period of time, a task that can be time and resource
consuming. One way of optimizing the lactose H2-BT is
decreasing the frequency and/or duration of breath sampling.
Thus, the aim of this study was to evaluate the sensitivity
and specificity of diagnosing lactase deficiency by measuring
H2 in fewer breath samples, whereas the results of the 3 h
H2-BT with 15-min sampling intervals served as gold standard.
baseline
0 min 15
This study is a retrospective analysis of prospectively
collected lactose H2-BT data from patients referred for lactose
intolerance testing to the Gastrointestinal Function Laboratory of the Division of Gastroenterology and Hepatology at
the University Hospital Zurich between 1999 and 2005. The
Ethics Committee (Institutional Review Board) approved the
retrospective analysis of the data collected as part of routine
clinical investigation.
Patients referred for lactose H2-BT were asked to be fasting
and refrain from smoking for at least 6 h before the test. The
appointment letter included instructions to discontinue
antibiotics at 4 weeks and laxatives at 1 day before testing.
Before the beginning of the test, two samples of mixed
end-expired air were aspirated into a 20 ml plastic syringe
(Injekt 20 ml, Braun, B Braun Medical AG, Emmenbrücke,
Switzerland) fitted with a three-way stopcock. The sample
with the higher amount of H2 was used as a baseline value,
and the test was performed provided both values were o20
parts per million (p.p.m.). After determining the baseline H2
breath concentration, the subject ingested 50 g of lactose
dissolved in 200 ml of water. Over a 3-h period, breath
samples were collected at 15-min intervals while the patient
was in sitting position. If the H2 concentration in at least two
breath samples was 420 p.p.m., the test was declared
positive (see below) and further breath sampling was
canceled. For the purpose of this study, lactose breath tests
lasting o2 h were not included in this analysis.
Breath H2 concentration was measured using a H2-BT
device (Stimotron Medizinische Geräte GmbH, Wendelstein,
Germany) equipped with an electrochemically working
hydrogen cell able to detect H2 concentrations in the range
of 0–250 p.p.m. with an accuracy of ±2% (1 p.p.m. at values
o50 p.p.m.). The instrument was calibrated using standardized compressed gas with a H2 concentration of 96.8 p.p.m.
(Calibration Gas, Stimotron, Wendelstein, Germany, compressed gas, no. 5, BDL. A/R). Air samples (20 ml each) were
injected in the H2-BT device and the H2 concentration was
read from a digital panel meter.
45
60
75
90
105 120
135 150 165 180 min
(X)
(A)
(B)
(C)
(D)
(E)
(F)
(G)
(H)
(I)
(J)
Materials and methods
30
Routine sample
Facultative sample
Figure 1 Standard lactose H2-breath testing with breath samples
collected every 15 min (scenario X). Facultative samples were taken
up to 180 min if the test was not positive by 120 min. Two breath
scenarios included samples collected at baseline, 60, 90 or 120 min
(scenarios A–C), three breath samples collected at baseline and
60 þ 90, 60 þ 120 or 90 þ 120 min, respectively (scenarios D–F), five
breath samples collected at baseline and every 15 min between
45–90, 60–105 or 75–120 min (scenarios G–I) and at baseline and
every 30 min (Figure 1; scenario J).
From the data collected at every 15 min over a period of 3 h
(standard test; Figure 1 scenario X) defined as ‘gold standard’,
we simulated 10 scenarios with breath samples collected after
longer time intervals and/or shorter test duration. Various
scenarios included two breath samples collected at baseline,
60, 90 or 120 min (Figure 1; scenarios A–C), three breath
samples collected at baseline and at 60 þ 90, 60 þ 120 or
90 þ 120 min, respectively (Figure 1; scenarios D–F), five breath
samples collected at baseline and at every 15 min between
45–90, 60–105 or 75–120 min (Figure 1; scenarios G–I), and
at baseline and at every 30 min (Figure 1; scenario J).
The standard (breath samples collected at every 15 min)
lactose H2-BT was considered positive if the H2 concentration in the exhaled air exceeded 20 p.p.m. above baseline at
least twice during the monitoring period (Metz et al., 1975).
For the abbreviated scenarios, the test was considered positive
if the H2 concentration in the exhaled air exceeded 20 p.p.m.
above baseline at least once.
Statistical analysis
Sensitivity, specificity, positive predictive value, negative
predictive value and positive and negative likelihood ratios
were calculated for different scenarios with either less
frequent sampling or shorter test-duration.
Results
Between July 1999 and December 2005, 1127 patients underwent a lactose H2-BT in the Department of Gastroenterology
European Journal of Clinical Nutrition
Diagnosing lactase deficiency in three breaths
M Oberacher et al
616
100%
80
100%
92%
# patients with >20 ppm H2breath concentration
66
90%
85%
80%
60
70%
53
70%
51
50
60%
54%
38
40
30
50%
24
35%
26
40%
30%
20
12
10
0
0
0
0
20%
15%
Cumulative % of patients
67
70
10%
4%
0%
15
30
45
60
75
Time (minutes)
90
105
120
Figure 2 Number of patients with an increase 420 p.p.m. in the H2-breath concentration at a given point in time (columns) and cumulative
percentage of patients with an 420 p.p.m. among the ones with a positive test result (line).
Table 1 Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and positive and negative likelihood ratios (LR þ and LR)
of two-, three- and five-breath sample scenarios
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
LR þ
LR
Baseline þ 60 min
Baseline þ 90 min
Baseline þ 120 min
52.5
81.9
92.6
100.0
99.7
99.2
100.0
99.3
98.1
81.7
92.1
96.6
N/A
291.6
109.9
0.475
0.182
0.075
Baseline þ 60 þ 90 min
Baseline þ 60 þ 120 min
Baseline þ 90 þ 120 min
83.4
95.0
95.0
99.7
99.2
98.9
99.3
98.2
97.6
92.7
97.6
97.6
296.8
112.7
84.5
0.167
0.051
0.051
Baseline þ Q15 min (45–90 min)
Baseline þ Q15 min (60–105 min)
Baseline þ Q15 min (75–120 min)
84.6
91.7
96.1
99.2
98.9
98.0
97.9
97.5
95.9
93.1
96.2
98.2
100.4
81.6
48.9
0.156
0.084
0.039
Baseline þ 30 þ 60 þ 90 þ 120 min
99.7
98.7
97.4
99.9
78.9
0.003
Scenario
and Hepatology of the University Hospital of Zurich.
Data from 78 patients were not included in the current
analysis, as the test was shorter than 2 h. In the 1049
patients fulfiling inclusion/exclusion criteria, a positive
result (that is, increase in excreted H2 concentration
420 p.p.m. above baseline in at least two samples collected
at 15 min time intervals) was found in 337 (32%) patients.
Figure 2 shows the number of patients with H2 concentration 420 p.p.m. and the cumulative percentage of patients
with H2 concentration 420 p.p.m. among those with
positive test result at each time point.
The sensitivity and specificity of two-sample tests ranged
from 53 to 93% and 99 to 100%, respectively. Three-breath
scenarios had better sensitivities (83–95%) and slightly lower
specificities (99–100%), whereas five-breath scenarios had
similar sensitivities (83–95%) and specificities (98–99%) to
the three-breath scenarios. Overall, the best sensitivity and
specificity was achieved by the five breath sample every
30 min scenario (sensitivity 99.7% and specificity 98.7%).
Sensitivity, specificity, positive predictive value, negative
European Journal of Clinical Nutrition
predictive value and positive and negative likelihood ratios
of individual two-, three- and five-breath sample scenarios
are summarized in Table 1.
Discussion
The results of this study, including over 1000 patients,
indicate that measuring breath H2 concentrations at hourly
intervals over a 2-h period has an excellent (495%)
sensitivity and specificity to diagnose lactase deficiency
when compared with standard 15-min sample breath testing.
On the other hand, shortening the duration of the measurement period to o2 h makes H2-breath testing for lactase
deficiency less sensitive. Thus, lactose H2 breath testing can
be simplified by decreasing the number of breath samples
but not by shortening the duration of the study to o2 h.
Studies evaluating options to simplify lactose H2-BTs have
been previously reported. Evaluating 132 consecutive lactose
H2-BTs with samples collected at every 30 min for 180 min,
Diagnosing lactase deficiency in three breaths
M Oberacher et al
617
Abramowitz et al. (1986) noted that all patients with a
positive test result had abnormal elevations of breath
H2 concentration at 120 min. When examining only the
120-min samples without subtracting the initial concentrations, 4/77 (5.2%) negative tests would have been falsely
positive. On the basis of these findings they concluded that
measuring breath H2 concentrations at 0 and 120 min were
sufficient to identify lactose malabsorption in 100% of patients
without a relevant loss of specificity.
The group of Ford et al. (2002) investigated ways to
simplify 3-h lactose H2-BTs with samples collected at 30 min
intervals, and found a high sensitivity (97%) of hourly
collected samples compared with the standard analysis.
Showing that longer intervals between breath-samplings
were not associated with an important loss of sensitivity, the
authors concluded that even in high-volume center using
the simplified test only one to two subjects per year would be
misdiagnosed.
Compared with our findings, Ford et al. performed the
standard test over a 3-h interval but challenged subjects with
only 25 g orally administered lactose. This is in accordance
with the recommendation that lower doses of lactose require
longer study durations (Solomons et al., 1980).
Other groups challenge the concept of short breath test
and argue for more frequent sampling and a longer test
duration. Casellas and Malagelada (2003) investigated the
sensitivity and specificity of a short H2-BT compared with a
5 h test as gold standard. They found that the best results
were obtained with a 3-h test when only 5% of studies were
false negative relative to a 5-h test. When the test was
shortened to o3 h, sensitivity dropped to 37%, values
regarded by these authors as unacceptable. To a similar
conclusion arrived, Rosado and Solomons (1983) who
postulated that the frequency and timing of breath collections constitute an important determinant of the sensitivity
and specificity of H2-BT. This was based on the results from
41 adult subjects who drank 360 ml cow’s milk and breath
samples were collected at 30-min intervals for 5 h indicating
that longer time intervals (that is, every hour or every other
hour) lead to a reduction in sensitivity to o90% while
maintaining a specificity of 100%. The most likely explanation could be the lower lactose load (360 ml milk contain
18 g lactose) used by (Rosado and Solomons, 1983 compared
with the 50 g lactose used in our study (that is, the
equivalent of over 1000 ml milk).
In a similar study design to ours, Di Camillo et al. (2006)
evaluated the implications of simplifying a 4-h 30-min
breath sampling lactose H2-BT in a large group of patients.
The authors concluded that the H2-BT could be effectively
performed with only three breath samples (0, 120 and
180 min) without a relevant loss of sensitivity (94.1%). On
the other hand, they argued that shortening the test to o3 h
would lead to a decline in sensitivity to an unacceptable low
level (about 70%). Although our study design was not geared
to test this hypothesis, it is of note that Di Camillo used a
low dose of lactose (0.5 g/kg body weight up to a maximum
of 25 g) a possible explanation for the low sensitivity of the
test in the first 2 h.
This study has some limitations. Several studies argued
that lactose H2-BT o3 h lose sensitivity. Still, the osmotic
activity of a high dose of lactose (50 g) most likely accelerates
intestinal transit thus increasing the likelihood of a positive
test in lactose deficient patients even before 2 h. The high
lactose load could, because of the hyperosmolarity lead to a
rapid orocoecal transit time, and thus increase of the overall
positivity rate. A recent consensus conference recommended
that a 20–25 g lactose load should preferred to the standard
50 g lactose load, but in either case a 10% iso-osmolar
solution should be used (Gasbarrini et al., 2009).
Even though the use of 50 g of lactose is challenged, it was
the standardized dose for many years allowing us to
accumulate such a large number of observations. Because
of the more physiological dosage of 25 g of lactose,
corresponding to 500 ml of milk, we suggest in future to
perform the challenge with only 25 g of lactose in an 10%
iso-osmolar solution. Second, the test applies mainly to
central European population, in which the lactase deficiency
is approximately 20–40%. In an attempt to allow a better
generalization of our results we reported positive and
negative likelihood ratios, as these values are independent
of the prevalence of a disease. Third, one might criticize that
we used a single value exceeding baseline by 20 p.p.m. to
declare the test positive. This cutoff value has been reported
by others (Veligati et al., 1994) who verified that a rise of
breath H2 concentration of X20 p.p.m. over baseline
correlated better than a rise X10 p.p.m. with subsequent
symptom development. This study takes into account only
the amount of exhaled H2 after ingestion of 50 g lactose
without evaluating symptoms developed during the study.
Although symptom assessment is an important part in the
clinical evaluation of patients suspected of having lactose
intolerance, the aim of this study was to evaluate the impact
of a simplified design of lactose H2-BT on the objective
results of this test, that is, the assessment of lactase
deficiency. Evaluating combination of both clinical features
and technical data would probably increase the test validity.
Still, the relevance of symptoms developed in response to
oral lactose challenge is controversial. Tolliver et al. (1994)
found that 52% of patients did not relate their symptoms
with oral intake of lactose. Suarez et al. (1995) argue that
when ingesting lactose containing or lactose-free milk in
blinded fashion patients with proven lactase deficiency
report similar intense symptoms. Vernia et al. (1995) found
that symptoms in lactose-intolerant patients persisted even
when being on a lactose-free diet arguing that another
underlying factors (for example, irritable bowel syndrome)
have a more important role in symptom genesis. The gold
standard used in this study (3-h lactose H2-BT) is influenced
by a variety of factors. Cigarette smoking induces an increase
in breath H2 concentrations due to exogenous contamination of alveolar gas with H2 from the cigarette (Rosenthal
and Solomons, 1983). Hyperventilation, teeth brushing,
European Journal of Clinical Nutrition
Diagnosing lactase deficiency in three breaths
M Oberacher et al
618
mouth wash and exercise have been shown to reduce the
breath hydrogen concentration (Thompson et al., 1985;
Ehrenpreis et al., 2002). Last but not least, lactose H2-BT
requires the presence of H2-producing flora which, according
to previous studies may not be present in up 20% of
individuals (Vogelsang et al., 1998).
Conclusion
In summary, the results of this study indicate that limiting a
50 g lactose H2-BT to three breaths (baseline, 60/90 and
120 min) maintains an excellent sensitivity and specificity to
diagnose lactase deficiency when compared with an extensive protocol measuring breaths every 15 min over a 2-h
period. Shortening the study to o2 h leads to an important
decline in sensitivity of the test.
In conclusion, 50 g lactose H2-BT with can be simplified to
include only 3 breaths samples over a 2 h period. Future
studies will show if the same simplification can be applied to
20–25 g lactose H2-BTs.
Conflict of interest
The authors declare no conflict of interest.
References
Abramowitz A, Granot E, Tamir I, Deckelbaum RJ (1986). Two-hour
lactose breath hydrogen test. J Pediatr Gastroenterol Nutr 5, 130–133.
Anh NT, Thuc TK, Welsh JD (1977). Lactose malabsorption in adult
Vietnamese. Am J Clin Nutr 30, 468–469.
Bayless TM, Rothfeld B, Massa C, Wise L, Paige D, Bedine MS (1975).
Lactose and milk intolerance: clinical implications. N Engl J Med
292, 1156–1159.
Casellas F, Malagelada JR (2003). Applicability of short hydrogen
breath test for screening of lactose malabsorption. Dig Dis Sci 48,
1333–1338.
Densupsoontorn N, Jirapinyo P, Thamonsiri N, Chantaratin S,
Wongarn R (2004). Lactose intolerance in Thai adults. J Med Assoc
Thai 87, 1501–1505.
Di Camillo M, Marinaro V, Argnani F, Foglietta T, Vernia P (2006).
Hydrogen breath test for diagnosis of lactose malabsorption: the
importance of timing and the number of breath samples. Can J
Gastroenterol 20, 265–268.
Ehrenpreis ED, Swamy RS, Zaitman D, Noth I (2002). Short duration
exercise increases breath hydrogen excretion after lactulose
ingestion: description of a new phenomen. Am J Gastroenterol
97, 2798–2802.
European Journal of Clinical Nutrition
Enattah NS, Sahi T, Savilahti T, Terwilliger JD, Peltonen L, Järvela I
(2002). Identification of a variant associated with adult-type
hypolactasia. Nat Genet 30, 233–237.
Farup PG, Monsbakken KW, Vandvik PO (2004). Lactose malabsorption
in a population with irritable bowel syndrome: prevalence and
symptoms. A case-control study. Scand J Gastroenterol 39, 645–649.
Ford DC, James C, Armstrong D (2002). Lactose hydrogen breath test
for the diagnosis of lactose intolerance: hourly breath samples
are adequate. Canadian Digestive Disease Week, Abstract, http://
www.pulsus.com/cddw2002/abs/abs161.htm; last accessed 09
November 2010.
Gasbarrini A, Corazza GR, Gasbarrini G, Montalto M, Di Stefano M,
Basilisco G et al (2009). Methodology and indications of H2-breath
testing in gastrointestinal diseases: the Rome Consensus Conference. Aliment Pharmacol Ther 29(Suppl 1), 1–49.
Gudmand-Hoyer E (1994). The clinical significance of disaccharide
maldigestion. Am J Clin Nutr 59, 735–741.
Metz G, Jenkins DJ, Peters TJ, Newman A, Blendis LM (1975).
Breath hydrogen as a diagnostic method for hypolactasia. Lancet
1, 1155–1157.
Newcomer AD, McGill DB, Thomas PJ, Hofmann AF (1975).
Prospective comparison of indirect methods for detecting lactase
deficiency. N Engl J Med 293, 1232–1236.
Pohl D, Savarino E, Hersberger M, Behlis Z, Stutz B, Goetze O et al.
(2010). Excellent agreement between genetic and hydrogen breath
tests for lactase deficiency and the role of extended symptom
assessment. Br J Nutr 19, 1–8.
Rosado JL, Solomons NW (1983). Sensitivity and specificity of the
hydrogen breathanalysis test for detecting malabsorption of
physiological doses of lactose. Clin Chem 29, 545–548.
Rosenthal A, Solomons NW (1983). Time-course of cigarette
smoke contamination of clinical hydrogen breath-analysis tests.
Clin Chem 29, 1980–1981.
Solomons NW (2002). Fermentation, fermented tools and lactose
intolerance. Eur J Clin Nutr 56(Suppl 4), S50–S55.
Solomons NW, Garcı́a-Ibáñez R, Viteri FE (1980). Hydrogen breath
test of lactose absorption in adults: the application of physiological
doses and whole cow’s milk sources. Am J Clin Nutr 33, 545–554.
Suarez FL, Savaiano DA, Levitt MD (1995). A comparison of
symptoms after the consumption of milk or lactose-hydrolyzed
milk by people with self-reported severe lactose intolerance. N Engl
J Med 333, 1–4.
Thompson DG, Binfield P, De Belder A, O’Brien J, Warren S, Wilson
M (1985). Extra intestinal influences on exhaled breath hydrogen
measurements during the investigation of gastrointestinal disease.
Gut 26, 1349–1352.
Tolliver BA, Herrera JL, DiPalma JA (1994). Evaluation of patients
who meet clinical criteria for irritable bowel syndrome. Am J
Gastroenterol 89, 176–178.
Veligati LN, Treem WR, Sullivan B, Burke G, Hyams JS (1994). Delta
10 ppm versus delta 20 ppm: a reappraisal of diagnostic criteria for
breath hydrogen testing in children. Am J Gastroenterol 89, 758–761.
Vernia P, Ricciardi MR, Frandina C, Bilotta T, Frieri G (1995). Lactose
malabsorption and irritable bowel syndrome. Effect of a long-term
lactose-free diet. Ital J Gastroenterol 27, 117–121.
Vogelsang H, Ferenci P, Frotz S, Meryn S, Gangl A (1998). Acidic
colonic microclimate-possible reason for false negative hydrogen
breath tests. Gut 29, 21–26.