EJINME-03306; No of Pages 8

European Journal of Internal Medicine xxx (2016) xxxxxx

Contents lists available at ScienceDirect

European Journal of Internal Medicine

journal homepage: www.elsevier.com/locate/ejim

Original Article

Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary

care setting: A large population study
Christian Selmer a,b,, Jesper Clausager Madsen c, Christian Torp-Pedersen e,
Gunnar Hilmar Gislason a,d,f, Jens Faber d,g
a

Department of Cardiology, Gentofte University Hospital, Hellerup, Denmark

Department of Endocrinology, Amager and Hvidovre University Hospital, Copenhagen, Denmark
Copenhagen General Practitioners Laboratory, Copenhagen, Denmark
d
Faculty of Health and Medical Sciences, University of Copenhagen, Denmark
e
Institute of Health, Science and Technology, Aalborg University, Aalborg, Denmark
f
National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark
g
Department of Endocrinology, Herlev University Hospital, Herlev, Denmark
b
c

a r t i c l e

i n f o

Article history:
Received 27 February 2016
Received in revised form 5 July 2016
Accepted 27 July 2016
Available online xxxx
Keywords:
Hyponatremia
Sodium
SIADH
Mortality
Neoplasm
Cancer

a b s t r a c t
Background: Hyponatremia has been associated with increased all-cause mortality in hospitalized individuals.
In this study we examine the risk of all-cause mortality in primary care subjects with hyponatremia, while also
exploring the association with subsequent diagnosis of cancer.
Methods: Retrospective cohort study on subjects who underwent blood tests, consulting their general practitioner
20002012 in Copenhagen, Denmark. Reference range for sodium was 135145 mmol/L, and mild, moderate, and
severe hyponatremia were dened as 130135, 125129, and b 125 mmol/L, respectively. Primary outcome was
all-cause mortality, and secondary outcomes overall and specic types of cancer diagnoses.
Results: Among 625,114 included subjects (mean age 49.9 [SD 18.4] years; 43.5% males), 90,926 (14.5%) deaths
occurred. All-cause mortality was increased in mild, moderate, and severe hyponatremia (age-adjusted mortality
rates [IRs, incidence rates] 26, 30, and 36 per 1000 person-years (py), respectively and incidence rate ratios [IRRs]
1.81 [95% CI: 1.761.85], 2.11 [2.002.21], and 2.52 [2.262.82], respectively) compared with individuals with
normonatremia (IR 14 per 1000 py). For the secondary endpoint an increased level-dependent risk was found
with lower sodium levels in relation to cancer overall, head and neck cancers, and pulmonary cancer, with severe
hyponatremia associated with the highest IRRs (1.77 [1.392.24], 5.24 [2.1712.63]), and 4.99 [3.497.15],
respectively).
Conclusions: All levels of hyponatremia are associated with all-cause mortality in primary care patients and
hyponatremia is linked to an increased risk of being diagnosed with any cancer, particularly pulmonary and
head and neck cancers.
2016 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.

1. Introduction
Hyponatremia in hospitalized patients is very common, and studies
on newly hospitalized patients have shown a strong association
between hyponatremia and an increased risk of death, as recently
reviewed in a meta-analysis on 81 studies including 850,222 patients
total (17.4% with hyponatremia) [1,2]. In this analysis, hyponatremia
was associated with 2.53.5 times increased mortality during the

Corresponding author at: Department of Cardiology, Research 1, Gentofte University

Hospital, Niels Andersens Vej 65, 2900 Hellerup, Denmark, Fax: +45 70201283.
E-mail address: cselmer@gmail.com (C. Selmer).

hospital stay largely independent of the underlying disease. However,

the majority of these studies were based on selected cohorts known to
have increased mortality, and most of them were relatively small. In a
recent Danish cohort study of 279,508 individuals with a rst-time
acute admission to a medical department, the 30 day and 1 year mortality were increased 1.7 and 1.4 times in hyponatremia, respectively [3].
It is unclear whether this association is caused by hyponatremia itself
or it is caused by increased vulnerability due to underlying medical
conditions or concomitant medications.
Hyponatremia is often seen among patients with active cancer
in 530% of patients [46]. When looking at subtypes of cancers, there
is an increased prevalence of hyponatremia especially in lung cancer,
but also upper respiratory tract and gastrointestinal cancers [7]. In both
untreated cancer and later during treatment, several chemotherapeutics,

http://dx.doi.org/10.1016/j.ejim.2016.07.028
0953-6205/ 2016 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

nausea, vomiting, and pain seem to induce hyponatremia. This is

most often due to the syndrome of inappropriate anti-diuretic hormone
(vasopressin) (SIADH) [8]. Thus, hyponatremia reects the presence
of active cancer tissue and is also a response to treatment. With the
recent development of selective vasopressin receptor antagonists for
treatment of hyponatremia, it is important to study the impact of
hyponatremia in patients evaluated in primary care [9,10].
Only a few studies have been carried out on primary care patients;
however, recently Hoorn et al. reported an increased risk of allcause mortality in 5208 elderly subjects, 399 with hyponatremia,
from the Rotterdam Study [11]. With the present large-scale study,
we evaluated if hyponatremia in the primary care setting was a predictor for all-cause mortality, as well as for the diagnosis of several
clinically overt cancer types including lung, gastrointestinal, and
urogenital. We performed extensive sensitivity analyses to examine
the temporal importance of hyponatremia, followed by spontaneous
normalization.

2.3. Co-morbidity and concomitant medical therapy

We identied co-morbidities from the Danish National Patient
Registry using the ICD-10 diagnosis codes for e.g. heart failure, adrenal
insufciency, and renal failure (Supplementary Table A) [19,20]. The
Charlson Comorbidity Index was based on 16 pre-specied diagnoses
up to ve years prior to cohort entry [21,22]. We identied all claimed
prescriptions of commonly used medications known to potentially
cause disturbances in serum sodium from the Danish Register of Medicinal Product Statistics [23].
2.4. Outcomes
The primary outcome was all-cause mortality. Secondary outcomes
included the diagnoses of head and neck, pulmonary, gastrointestinal,
urogenital, hematological, and other cancers.
2.5. Statistical analysis

2. Materials and methods

2.1. Study setting
In Denmark, each resident has a permanent and unique civil registration number, enabling individual level-linkage between nationwide
administrative registers on healthcare usage [12]. Since 1978, the
Danish National Patient Registry has registered all hospital contacts
in Denmark [13]. Each admission is registered with one primary and,
if appropriate, one or more secondary diagnoses according to the
World Health Organization International Classication of Diseases
10th revision (ICD-10). The Danish Register of Medicinal Product
Statistics holds information regarding all claimed prescriptions (coded
according to the international Anatomical Therapeutic Chemical [ATC]
classication) in Denmark since 1995 [14]. The registry also includes
information on the date of dispensation, strength, and quantity dispensed. Due to partial reimbursement of drug expenses by the Danish
healthcare authorities, all pharmacies are required to provide information that ensures complete and accurate registration [15]. Vital status
was obtained from the Central Population Register, which records all
deaths within 14 days [12]. Specic causes of death were obtained
from the Danish Register of Causes of Death [16]. Annual incomes
were retrieved from the Integral Database for the Danish Labour Market
[17], and socioeconomic status was dened by the average yearly gross
household income in a 5-year period prior to inclusion in the study.
We conducted and reported this study following the Strengthening
the Reporting of Observational Studies in Epidemiology (STROBE)
Initiative guidelines [18].
2.2. Population
The study cohort comprised citizens of Copenhagen (the capital
of Denmark), aged 18 years or older, who had serum sodium analyzed
in the period from 1 January 2000 to 31 December 2012. Each subject
entered the cohort at the rst known assessment of serum sodium
and was followed until 31 December 2012, migration, or death. The
Copenhagen General Practitioners Laboratory analyses tests from
primary care physicians in the Copenhagen area, except for the municipality of Frederiksberg. This corresponded to approximately 1.17 million
inhabitants in 2009. Serum sodium concentrations were determined
using the commercially available ADVIA Centaur System (Bayer/
Siemens, Tarrytown, NY), according to the instructions of the manufacturer. Individuals where categorized according to their sodium levels
at the time of rst testing into three levels of hyponatremia. The normal
reference range for serum sodium was 135145 mmol/L, and mild, moderate, and severe hyponatremia were dened as 130135, 125129, and
b125 mmol/L, respectively.

Baseline characteristics are presented as numbers with percentages

for categorical variables and as means (SD) for continuous variables.
Median follow-up time was reported with the interquartile range
(IQR). Incidence rates (IRs) were the number of events per 1000
person-years (py), stratied by sodium-level. We constructed ageadjusted Kaplain Meier curves for the primary endpoint of all-cause
mortality. We assessed cumulative incidence proportion curves for
any cancer at all levels of hyponatremia using a competing risk model
to adjust for all-cause mortality in the cohort [24]. We constructed
time-dependent Poisson regression models to estimate incidence rateratios (IRRs, with 95% condence intervals [CIs]) for each study
outcome. The Poisson regression models were adjusted for age, sex,
and calendar year; therefore, they included two time scales: Calendar
time with bands split in 1-year periods after 1st of January 2000, and
duration time since rst serum sodium measurement. Age was calculated at the beginning of each interval. Individuals were censored at the
time of death, the end of the follow-up period (31st of December
2012), or at migration. The signicance level was set at 5% in all analyses, including testing for interactions.
We validated our primary ndings with several sensitivity analyses. First, we stratied all results by sex and age. Second, we stratied
whether a second serum sodium testing demonstrated a normalized
or a continued state of hyponatremia. Third, we adjusted for baseline
covariates including medication. Fourth, we limited the follow-up
time to only 6 months to assess the short-term outcome following a
single measurement of serum sodium. Within this 6-month follow-up
period, we also limited the cohort to patients treated with diuretics
only, patients treated with Selective Serotonin Reuptake Inhibitors
(SSRI) only, and patients treated with neither diuretics nor SSRI. Finally,
we looked at the full follow-up period under three special circumstances: 1) including only SSRI treated individuals with no diuretics
use; 2) introducing a 30-day grace period (postponing the follow-up
period by 30 days) in the analysis to assess the potential issue of
confounding by indication; and 3) adjusting the main model for the
Charlson Comorbidity Index and use of SSRI and diuretics; 4) adjusting
for hypothyroidism at time of sodium evaluation dened as thyroidstimulating hormone (TSH) N4.5; 5) regrouping hyponatremia using
frequency groups. All statistical analyses were performed with the
SAS Statistical Software package version 9.2 (SAS Institute Inc., Gary,
NC, USA) and Stata Software version 11 (StataCorp, College Station,
TX, USA).
3. Results
A total of 625,114 individuals were included (Fig. 1) at the rst
serum sodium testing from 2000 to 2012. Baseline characteristics of
the cohort are presented in Table 1. The study cohort comprised more

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

Fig. 1. Flowchart of study cohort selection.

females (56.5%) than males and the average age was 49.9 ( 18.4)
years. Individuals with serum sodium within the normal reference
range had a low degree of comorbidity, and the frequency of comorbidities increased with lower levels of serum sodium. During a total followup time of 3,907,925 py (median follow-up 6.2 [IQR: 3.19.4] years),
90,926 (14.5%) subjects died.
3.1. All-cause mortality
The unadjusted incidence-rates (IRs) of the primary endpoint of allcause mortality were increased for all subjects with hyponatremia
(Table 2). Subjects with lower levels of serum sodium had the highest
adjusted rates, i.e. when estimating IRs for subjects at age 65 years.
The KaplanMeier curves for the primary endpoint of all-cause mortality, stratied by levels of serum sodium and adjusted for age at 65 years,
are shown in Fig. 2A. Results from the Poisson regression analysis,

adjusted for age, sex, and calendar-year, are illustrated in Fig. 3. There
was a signicant level-dependent increased risk of all-cause mortality
for lower serum sodium levels.
Results for the sensitivity analysis are presented in Fig. 4 and Supplementary Fig. A. Overall, there were no differences from the main model
when adjusting for Charlson Comorbidity Index and important covariates, such as the use of SSRI and diuretics; although, the numerical
values of the risk estimates were slightly lower. On the other hand,
males and especially individuals aged b65 years had substantially
higher risk estimates. Importantly, the risk of all-cause mortality continued to increase when serum sodium had normalized on a second blood
test, and a 30-day grace period did not change the overall results from
the main model. Limiting the follow-up time to only 6 months did
signicantly increase the risk estimates, but limiting the cohort to
diuretic and/or SSRI users did not change the overall ndings. Interestingly, the estimated IRR was very high among individuals with no

Table 1
Baseline characteristics of the cohort.
Normal sodium
(S-Na 135145 mmol/L)

Mild hyponatremia
(S-Na 130135 mmol/L)

Moderate hyponatremia
(S-Na 125129 mmol/L)

Severe hyponatremia
(S-Na b 125 mmol/L)

n = 610,597

n = 11,757

n = 2304

n = 456

Age in years
Sex (Male)
Serum sodium (mmol/L)

49.54 18.27
266,244 (43.6%)
140.65 2.12

67.32 16.69
4539 (38.6%)
132.64 1.33

69.15 15.42
906 (39.3%)
127.63 1.30

67.99 15.38
189 (41.4%)
121.07 3.35

Diagnoses
Adrenal insufciency
Renal disease
Liver disease
Heart failure
Diabetes

625 (0.1%)
4325 (0.7%)
6724 (1.1%)
14,287 (2.3%)
3561 (0.6%)

34 (0.3%)
207 (1.8%)
501 (4.3%)
1052 (8.9%)
342 (2.9%)

13 (0.6%)
36 (1.6%)
136 (5.9%)
233 (10.1%)
72 (3.1%)

b3 (0.4%)
16 (3.5%)
39 (8.6%)
49 (10.7%)
19 (4.2%)

Charlson comorbidity index

0
1
2
3

595,188 (97.5%)
11,690 (1.9%)
1568 (0.3%)
2151 (0.4%)

10,435 (88.8%)
928 (7.9%)
152 (1.3%)
242 (2.1%)

1976 (85.8%)
229 (9.9%)
43 (1.9%)
56 (2.4%)

376 (82.5%)
52 (11.4%)
9 (2.0%)
19 (4.2%)

Medication
Diuretics
Thiazide
Loop
SSRIs
NSAIDs
Paracetamol

81,185 (13.3%)
58,591 (9.6%)
23,560 (3.9%)
49,645 (8.1%)
90,961 (14.9%)
68,804 (11.3%)

5560 (47.3%)
3939 (33.5%)
1816 (15.5%)
1832 (15.6%)
2292 (19.5%)
3576 (30.4%)

1352 (58.7%)
983 (42.7%)
401 (17.4%)
468 (20.3%)
465 (20.2%)
791 (34.3%)

281 (61.6%)
200 (43.9%)
93 (20.4%)
89 (19.5%)
105 (23.0%)
174 (38.2%)

Age and serum sodium are the mean standard deviation (SD). All other covariates are numbers (%).

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

Table 2
Rates and incidence rate ratios (IRR) of all-cause mortality, pulmonary, gastrointestinal, and urogenital cancers.
Serum sodium

IRR

Events during
follow-up

Person-years

Incidence rate
(n/1000 py)

Adjusted incidence rate

(n/1000 py)

All-cause mortality
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

2.52 (2.262.82)
2.11 (2.002.21)
1.81 (1.761.85)
1 (11)

315
1559
6724
82,328

2.2
12.3
68.1
3825

142.9
126.3
98.8
21.5

36.2
30.1
25.8
14.1

Cancer overall
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

1.77 (1.392.24)
1.31 (1.171.47)
1.32 (1.261.39)
1 (11)

69
297
1527
32,339

2.1
11.8
65.3
3746

32.5
25.1
23.4
8.6

16.8
12.5
12.3
8.6

Head and neck cancer

b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

5.24 (2.1712.63)
3.35 (2.15.36)
2.52 (1.983.22)
1 (11)

5
18
71
1050

2.2
12.3
68
3822

2.3
1.5
1
0.3

1.3
0.9
0.6
0.2

Pulmonary cancer
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

4.99 (3.497.15)
2.18 (1.742.74)
1.87 (1.672.09)
1 (11)

30
76
333
5047

2.2
12.3
67.8
3820

13.7
6.2
4.9
1.3

6.6
2.9
2.4
1.2

Gastrointestinal cancer
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

1.27 (.642.55)
1.35 (1.021.78)
1.14 (.991.31)
1 (11)

8
50
214
4721

2.2
12.3
67.8
3815

3.6
4.1
3.2
1.2

1.4
1.5
1.2
1

Urogenital cancer
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

.48 (.181.28)
.61 (.42.88)
.94 (.821.07)
1 (11)

4
29
231
7343

2.2
12.3
67.5
3803

1.8
2.4
3.4
1.9

0.7
0.9
1.3
1.3

Hematologic cancer
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

1.83 (.694.89)
1.18 (.711.97)
1.58 (1.291.94)
1 (11)

4
15
102
1727

2.2
12.3
68
3822

1.8
1.2
1.5
0.5

0.8
0.5
0.7
0.4

Other cancers
b125 mmol/L
125129 mmol/L
130135 mmol/L
135145 mmol/L (ref.)

1.04 (.661.66)
1.23 (1.031.47)
1.26 (1.161.37)
1 (11)

18
122
639
14,239

2.2
12.1
66.6
3783

8.2
10.1
9.6
3.8

5.2
6.1
6.2
4.6

CI; Condence Interval. Py; person-years. Risk-estimates are adjusted for sex, age, and calendar-year. Age-adjusted incidence rates are adjusted to 65 years.

known diagnosis or medical prescriptions at baseline. In the sensitivity analysis focusing on the full follow-up period and narrowing
in on only SSRI treated individuals, introducing a 1-month grace
period, adjusting for concomitant hypothyroidism in a subgroup
of 442,560 individuals of whom 17,832 had a TSH N 4.5 mIU/L, as
well as regrouping individuals by frequency all yielded the same
overall results.
3.2. Subsequent diagnosis of cancer
For the secondary endpoints of subsequent diagnosis of cancer
overall, head and neck, pulmonary, gastrointestinal, and hematologic
cancers, the unadjusted and adjusted IRs were increased for all subjects
with hyponatremia, with increasing IRs at lower levels of sodium
(Table 2). The 12-month cumulative incidence of cancer overall
increased with the level of hyponatremia (Fig. 2B). Adjusted Poisson
regression analysis showed a level-dependent increased risk with the
lower serum sodium level in relation to subsequent cancer overall,
head and neck, and pulmonary cancers (Fig. 3). There was also a signicantly increased risk in subjects with moderate hyponatremia
(125129 mmol/L) with regard to gastrointestinal cancer, mild hyponatremia (130135 mmol/L) for hematologic cancer, and both mild

and moderate hyponatremia in relation to other cancers (but no clear

level dependent association).

4. Discussion
In this large cohort study of 625,114 primary care patients, we examined the association between rst found hyponatremia in 14,517 individuals and the risk of all-cause mortality and subsequent diagnosis of
cancer. We found that the risk of death increased in a level-dependent
fashion, with the highest risk in subjects with the lowest levels of
serum sodium. Compared with individuals with normal serum sodium,
mild, moderate, and severe hyponatremia were associated with 81%,
111%, and 152% increased risks of all-cause mortality, respectively.
These risks were not substantially altered after extensive sensitivity
analyses, which also highlighted that people aged b 65 years had an
especially high risk of all-cause mortality with a 9-fold increased risk
in those with severe hyponatremia. For the secondary endpoint, the
risk of any cancer diagnosis was signicantly increased during followup, with a 5-fold increased risk of being diagnosed with head and
neck cancer or pulmonary cancer following a laboratory test with severe
hyponatremia. To our knowledge, this is the rst report demonstrating

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

Fig. 2. (A) Hyponatremia and all-cause mortality: Age-adjusted KaplanMeier curves stratied by levels of serum sodium (adjusted to an age of 65-years). (B): Hyponatremia and
subsequent diagnosis of cancer overall. Cumulative incidence adjusted for the competing risk of all-cause mortality.

hyponatremia predicting a cancer diagnosis in a large unselected cohort

of subjects in a primary care setting.
4.1. Mortality
Most previous studies concerning hyponatremia and mortality have
been on hospitalized patients and only a few prospective studies exist
on community dwelling subjects. The Copenhagen Holter Study on
671 community dwelling subjects (76 with hyponatremia) reported
an overall 2-fold increased risk of a combined endpoint of myocardial
infarction and all-cause mortality [25]. However, that study was not
able to adjust for a number of important comorbidities and concomitant
medications. Hoorn et al. in the Rotterdam Study investigating community dwelling elderly people, found at baseline an increased history of
recent falls and increased use of diuretics in people with hyponatremia
versus normonatremia [11]. Furthermore, during 67 years of followup, they observed a 21% increase in all-cause mortality as well as a
39% increased risk of incident non-vertebral fractures [11]. However,
hyponatremia was not associated with lower bone mass density
(BMD) in that study. With regard to the increase of falls induced by
hyponatremia, Renneboog et al. in a casecontrol study of 122 elderly
patients admitted to a medical emergency department found that
hyponatremia could be an important risk factor for falls due to gait
and attention impairment [26]. Verbalis et al. suggested that hyponatremia is a risk factor for development of osteoporosis based on
epidemiological data, and they demonstrated in animal studies that

hyponatremia leads to osteoclast activation, and thus enhanced bone

break down [27]. Thus, hyponatremia might be regarded as a risk factor
for increased falls due to gait instability and osteoporosis, especially in
the elderly [28]. However, it is unclear if hyponatremia itself causes an
increased risk of death or if the condition is simply a marker of concomitant medication or disease states causing changes in serum sodium.
In the primary care setting, four common confounders often occur:
Use of loop diuretics and potassium sparing diuretics causing natriuresis, use of SSRI, malignant conditions causing SIADH, and diabetes
mellitus causing hyperglycemia, and altered vasopressin metabolism
[2931]. Indeed, we found that the lower the serum sodium levels,
the more drugs were used, including diuretics (increasing from 13%
in normonatremia to 47%59% in the hyponatremic groups) and SSRI
(increasing from 8% to 16%20% in the hyponatremic groups). Nevertheless, adjusting for all these factors as well as other important comorbidities using the Charlson Comorbidity Index, and including liver
cirrhosis and congestive heart failure (conditions with an especially
high prevalence of hyponatremia) did not change the overall ndings
[32,33]. Of special interest were the individuals with no known baseline
diagnosis or medical use who seemed to have among the highest
IRR of death (4.3 in mild, 8.6 in moderate, and 14.5 in severe
hyponatremia, respectively). Our sensitivity analyses further elucidated
that people aged b 65 years had an IRR 34 times higher than those aged
N65 years. These data combined with the demonstration of a clear
doseresponse association with decreasing levels of serum sodium, an
association that was also found after excluding diuretic and SSRI users,

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

of hyponatremia on cancer outcomes among hospitalized patients [6,7].

We present the rst large-scale study on different levels of hyponatremia and subsequent diagnoses of major cancer subtypes in a
primary care setting. In individuals with severe hyponatremia, the
unadjusted incidence of any cancer diagnosis was increased 4-fold
(2-fold increased when adjusted for sex and age) with an adjusted
77% increased risk in the Poisson model. Interestingly, we demonstrated
a level-dependent 2 to 5-fold increased risk of pulmonary and head and
neck cancers in subjects with mild to severe hyponatremia. There was
also an increased risk for gastrointestinal, hematological, and other cancer types (including breast cancer) but with a less clear level-dependent
association. Since SIADH is most common in small-cell lung and head
and neck cancers, this nding may reect the actual tumor burden.
With regard to SIADH in head and neck cancers, case studies have
reported years of constant hyponatremia before a nal diagnosis was
reached, typically of low-grade malignant tumors [40]. The associations
between all levels of hyponatremia, particularly severe hyponatremia,
and a subsequent diagnosis of cancer underline the importance of
thoroughly evaluating unexplained hyponatremia with special focus
on pulmonary and head and neck cancers [41].
4.3. Strengths and limitations

Fig. 3. Incidence Rate Ratios (IRRs) of all-cause mortality, and diagnoses of pulmonary,
gastrointestinal, and urogenital cancer adjusted for age, sex and calendar year.

might indicate that hyponatremia itself is a causal factor. Although the

reason for this adverse effect of hyponatremia remains unknown,
it may be due to a combination of increased fall rates, fracture risk, or
potential direct adverse effects of low serum sodium on the cardiovascular system [34,35].
Interestingly, we demonstrated that the risk of death remained
increased even after serum sodium levels normalized on a second measurement. This normalization might indicate that the hyponatremia was
addressed by treatment of a specic condition or a change in medication. Whatever the reason, our data suggest even a single and isolated
reduced serum sodium level should be considered with care, since it
might signal a previously unrecognized clinically vulnerable situation.
To address the isolated effect of hyponatremia on morbidity and mortality, future studies need to assess the potential pathophysiological mechanisms associated with hyponatremia, and trials assessing the impact of
restoring normonatremia are urgently needed. A relatively new class of
drugs, vaptans, could be used in this setting, since they create aquaresis,
i.e. electrolyte free diuresis, and hence result in an increase of serum
sodium [36].
4.2. Subsequent diagnosis of cancer
Asymptomatic hyponatremia was described as a rst sign of pulmonary cancer in case reports [37,38]. Hyponatremia has even been suggested as a potential marker for screening for pulmonary cancer [39].
However, most of the epidemiological studies have focused on the prevalence of hyponatremia in different cancer subtypes (~30%) or the effect

This study's main strength is the large cohort of subjects from primary care who had serum sodium measured and had complete
follow-up data available. Around half of the population in the capital
region of Denmark had serum sodium measured and were included.
In general, observational studies should not make conclusions on causal
relationships, and that's why no nal conclusion can be made based
on this study as to whether hyponatremia itself is associated with
increased mortality or conditions causing hyponatremia explain this
association. The study design could be prone to selection bias, because
we did not know the specic reason for individuals having blood testing. However, usually serum sodium is measured as part of a larger
blood-testing package, and very seldom is it measured specically.
This makes serum sodium well suited for the kind of analyses we
have performed, reducing the risk of confounding by indication.
We did not have information about important clinical parameters,
e.g. body weight, blood pressure, serum lipid levels, echocardiographic
or electrocardiographic ndings, or smoking status. However, adjustment for socioeconomic status is a valid proxy for smoking, and
the Charlson Comorbidity Index that includes chronic obstructive
pulmonary disease (COPD) did not alter the overall results. Further,
the inclusion of data on all types of medication provided valid data for
patients followed exclusively in an outpatient setting, who did not
have a registered diagnosis in the Danish National Patient Registry.
This ensured that prevalent diagnoses such as diabetes and hypertension were included in the Charlson Comorbidity Index. Subjects were
categorized with mild, moderate, and severe hyponatremia based on
the results of their rst serum sodium measurement. It is possible that
long-standing hyponatremia or correction could be of importance;
although, our sensitivity analysis that stratied individuals with
normalized serum sodium did not change the overall results. However,
importantly to remember is that we did not have any data on specic
actions taken in order to correct hyponatremia such as treatment with
intravenous saline or vasopressin receptor antagonists. The Danish
population comprises mainly Caucasians, and extrapolation of the
results to other ethnic groups should be done with care. Likewise, as
the study included individuals in primary care settings, extrapolation
of our results to patients from in- or outpatient clinics should be performed with caution.
5. Conclusion
All-cause mortality increased in a level-dependent manner with the
increasing severity of hyponatremia among primary care patients.

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

Fig. 4. Incidence rate ratios (IRRs) of primary endpoint all-cause mortality stratied by gender, age group, and second lab test and adjusted for baseline covariates.

Furthermore, all levels of hyponatremia were associated with increased

risk of being diagnosed with any cancer, particularly pulmonary and
head and neck cancers. These results support that nding even mild
hyponatremia in the primary care setting could be serious and that
underlying conditions, including medication, should be identied. Further studies are needed to evaluate if the correction of hyponatremia
is associated with improved outcome.
Contributors
JF conceived the study and together with CS made primary contributions to the data collection and analysis, interpretation of results, and
writing the manuscript. GHG and CT-P contributed to the study conception and design. All authors contributed to the interpretation of results,
revised the manuscript critically for important intellectual content, and
approved the nal manuscript. CS is the guarantor.
Funding
The work was supported by an unrestricted grant from Otsuka
Pharma Scandinavia AB.
Ethical approval
This study was approved by The Danish Data Protection Agency
(Ref. No. 2007-58-0015/GEH-2014-018; I-Suite No. 02736). Retrospective register studies do not require ethical approval in Denmark. CS

and GHG had full access to the data and take full responsibility for its
integrity.
Conict of interest statement
GHG is supported by an unrestricted research scholarship from the
Novo Nordisk Foundation. CS has received non-CME-related fees from
Otsuka Pharma Scandinavia AB. JF has served as a consultant and
received non-CME-related fees from Otsuka.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.ejim.2016.07.028.
References
[1] Schrier RW, Verbalis JG. The syndrome of inappropriate antidiuretic hormone secretion and other hypoosmolar disorders. Diseases of the kidney and urinary tract.
Lippincott Williams & Wilkins; 2007.
[2] Corona G, Giuliani C, Parenti G, et al. Moderate hyponatremia is associated with increased risk of mortality: Evidence from a meta-analysis. PLoS One 2013;8:e80451.
[3] Holland-Bill L, Christiansen CF, Heide-Jrgensen U, et al. Hyponatremia and mortality
risk: A Danish cohort study of 279508 acutely hospitalized patients. Eur J Endocrinol
2015;173:7181.
[4] Hawkins RC. Age and gender as risk factors for hyponatremia and hypernatremia.
Clin Chim Acta 2003;337:16972.
[5] Zilberberg MD, Exuzides A, Spalding J, et al. Epidemiology, clinical and economic
outcomes of admission hyponatremia among hospitalized patients. Curr Med Res
Opin 2008.

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028

C. Selmer et al. / European Journal of Internal Medicine xxx (2016) xxxxxx

[6] Waikar SS, Mount DB, Curhan GC. Mortality after hospitalization with mild, moderate, and severe hyponatremia. Am J Med 2009;122:85765.
[7] Doshi SM, Shah P, Lei X, Lahoti A, Salahudeen AK. Hyponatremia in hospitalized
cancer patients and its impact on clinical outcomes. Am J Kidney Dis 2012;59:2228.
[8] Berghmans T. Hyponatremia related to medical anticancer treatment. Support Care
Cancer 1996;4:34150.
[9] Schrier RW, Gross P, Gheorghiade M, et al. Tolvaptan, a selective oral vasopressin
V2-receptor antagonist, for hyponatremia. N Engl J Med 2006;355:2099112.
[10] Corona G, Giuliani C, Verbalis JG, Forti G, Maggi M, Peri A. Hyponatremia
improvement is associated with a reduced risk of mortality: Evidence from a
meta-analysis. PLoS One 2015;10:e0124105.
[11] Hoorn EJ, Rivadeneira F, van Meurs JBJ, et al. Mild hyponatremia as a risk factor for
fractures: The Rotterdam study. 2011;26:18228.
[12] Olesen JB, Gislason GH, Charlot MG, et al. Calcium-channel blockers do not alter the
clinical efcacy of clopidogrel after myocardial infarction: A nationwide cohort
study. J Am Coll Cardiol 2011;57:40917.
[13] Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J
Public Health 2011;39:303.
[14] WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC
classication and DDD assignment. http://www.whocc.no/atc_ddd_publications/
guidelines/; 2012.
[15] Kildemoes HW, Sorensen HT, Hallas J. The Danish National Prescription Registry.
Scand J Public Health 2011;39:3841.
[16] Helweg-Larsen K. The Danish register of causes of death. Scand J Public Health 2011;
39:269.
[17] Baadsgaard M, Quitzau J. Danish registers on personal income and transfer
payments. Scand J Public Health 2011;39:1035.
[18] von Elm E, Altman DG, Egger M, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. Lancet 2007;370:14537.
[19] Madsen M, Davidsen M, Rasmussen S, Abildstrom SZ, Osler M. The validity of the
diagnosis of acute myocardial infarction in routine statistics: A comparison of mortality and hospital discharge data with the Danish MONICA registry. J Clin Epidemiol
2003;56:12430.
[20] Krarup LH, Boysen G, Janjua H, Prescott E, Truelsen T. Validity of stroke diagnoses in
a national register of patients. Neuroepidemiology 2007;28:1504.
[21] Nuttall M, van der Meulen J, Emberton M. Charlson scores based on ICD-10 administrative data were valid in assessing comorbidity in patients undergoing urological
cancer surgery. J Clin Epidemiol 2006;59:26573.
[22] Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sorensen HT. The predictive
value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of patients. BMC Med Res
Methodol 2011;11:83.
[23] Liamis G, Milionis H, Elisaf M. A review of drug-induced hyponatremia. Am J Kidney
Dis 2008;52:14453.

[24] Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing
risk. J Am Stat Assoc 1999.
[25] Sajadieh A, Binici Z, Mouridsen MR, Nielsen OW, Hansen JF, Haugaard SB. Mild
hyponatremia carries a poor prognosis in community subjects. 2009;122:67986.
[26] Renneboog B, Musch W, Vandemergel X, Manto MU, Decaux G. Mild chronic
hyponatremia is associated with falls, unsteadiness, and attention decits. 2006;
119:71.e718.
[27] Verbalis JG, Barsony J, Sugimura Y, et al. Hyponatremia-induced osteoporosis. 2010;
25:55463.
[28] Miriam Rachel Usala G, Fernandez SJ, Mete M, et al. Hyponatremia is associated with
increased osteoporosis and bone fractures in a large U. S. health system population.
J Clin Endocrinol Metab 2015:jc20151261.
[29] Friedman E, Shadel M, Halkin H, Farfel Z. Thiazide-induced hyponatremia. Reproducibility by single dose rechallenge and an analysis of pathogenesis. Ann Intern Med
1989;110:2430.
[30] Hillier TA, Abbott RD, Barrett EJ. Hyponatremia: Evaluating the correction factor for
hyperglycemia. Am J Med 1999;106:399403.
[31] Bankir L, Bardoux P, Ahloulay M. Vasopressin and diabetes mellitus. Nephron 2001;
87:818.
[32] Schou M, Valeur N, Torp-Pedersen C, Gustafsson F, Kber L. Plasma sodium and mortality risk in patients with myocardial infarction and a low LVEF. Eur J Clin Investig
2011;41:123744.
[33] Peterson JC, Paget SA, Lachs MS, Reid MC, Charlson ME. The risk of comorbidity.
2012;71:6357.
[34] Holm JP, Amar AOS, Hyldstrup L, Jensen JEB. Hyponatremia, a risk factor for osteoporosis and fractures in women. Osteoporos Int 2015;1-13.
[35] Mouallem M, Friedman E, Shemesh Y, Mayan H, Pauzner R, Farfel Z. Cardiac conduction defects associated with hyponatremia. Clin Cardiol 1991;14:1658.
[36] Robertson GL. Vaptans for the treatment of hyponatremia. Nat Rev Endocrinol 2011;
7:15161.
[37] Kamoi K, Kurokawa I, Kasai H, Mizusawa A, Ebe T. Asymptomatic hyponatremia due
to inappropriate secretion of antidiuretic hormone as the rst sign of a small cell
lung cancer in an elderly man. Intern Med 1998.
[38] Gschwantler M, Weiss W. Hyponatremic coma as the rst symptom of a small cell
bronchial carcinoma. Dtsch Med Wochenschr 1994;119:2614.
[39] Kasi PM. Proposing the use of hyponatremia as a marker to help identify high risk
individuals for lung cancer. Med Hypotheses 2012;79:3278.
[40] Sejling A-S, Thorsteinsson A-L, Pedersen-Bjergaard U, Eiken P. Recovery from
SIADH-associated osteoporosis: A case report. J Clin Endocrinol Metab 2014;
99(10):352730.
[41] Verbalis JG, Goldsmith SR, Greenberg A, et al. Diagnosis, evaluation, and treatment of
hyponatremia: expert panel recommendations. Am J Med 2013:S142.

Please cite this article as: Selmer C, et al, Hyponatremia, all-cause mortality, and risk of cancer diagnoses in the primary care setting: A large
population study, Eur J Intern Med (2016), http://dx.doi.org/10.1016/j.ejim.2016.07.028