Thiazide-Associated Hyponatremia

A Population-Based Study

Eline M. Rodenburg, MD; Ewout J. Hoorn, MD, PhD; Rikje Ruiter, MD, PhD; Jan J. Lous, MSc; Albert Hofman, MD, PhD; André G. Uitterlinden, PhD; Bruno H. Stricker, PhD; Loes E. Visser, PharmD, PhD


Am J Kidney Dis. 2013;62(1):67-72. 

In This Article



This study was embedded in the Rotterdam Study, an ongoing prospective population-based cohort study of chronic diseases in elderly Caucasians. All inhabitants of Ommoord, a suburb of the city of Rotterdam in the Netherlands, 55 years or older were invited in 1990 to participate in the study. The medical ethics committee of the Erasmus Medical Center approved the study and informed consent was obtained from all participants. The rationale and design of the study were described elsewhere.[18] The first cohort encompassed 7,983 individuals who were interviewed and examined at baseline in 1990–1993 (Rotterdam Study-1 [RS-I]). In 2000, all inhabitants of Ommoord 55 years and older at that time and not yet participating in RS-I were invited to participate in the extended cohort (RS-II). This cohort encompassed 3,011 individuals who entered the study after giving consent. In 2006, a further extension of the cohort (second extended cohort, RS-III) was initiated, in which 3,932 individuals 45 years and older were included.

Follow-up examinations are carried out periodically, while all participants are monitored continuously for major morbidity and mortality through linkage with general practitioner and municipality records. All available serum sodium levels of participants from the study population were gathered from a general practitioner's laboratory serving the area of the Rotterdam Study. Data were available for May 1997 up to March 2010.

Drug exposure has been monitored continuously since January 1, 1991, through computerized pharmacy records of the pharmacies in the Ommoord district. The pharmacy data include the Anatomical Therapeutical Chemical code, dispensing date, total amount of drug units per prescription, prescribed daily number of units, and product name of the drugs.


We followed up all participants from RS-I, RS-II, and RS-III from May 1997 until January 2008 because follow-up data for survival status were complete up to that date. Participants who were referred by their general practitioner to the laboratory for serum sodium level assessment were defined as having hyponatremia if they had a serum sodium level ≤135 mmol/L. Mild hyponatremia was defined as serum sodium level ≥130 and ≤135 mmol/L; moderate hyponatremia, as >125 and <130 mmol/L; and severe hyponatremia, as ≤125 mmol/L. The date of first hyponatremia (either mild or severe) episode was defined as the index date. For assessment of severe hyponatremia, the index date was the date of the first severe hyponatremia episode.

Drug Exposure

The exposure of interest was defined as use of a thiazide diuretic (alone or in combination) or related agent (chlorthalidone, mefruside, and indapamide) at the day of sodium measurement. The exposure period started at the prescription filling date and was calculated by dividing the number of units issued per prescription by the prescribed daily number of units. Participants who had hyponatremia on a date within an exposure period were considered as current users.

Daily doses were expressed as the daily number of standardized defined daily doses (DDD) according to a World Health Organization website.[19] For instance, the DDD for hydrochlorothiazide is 25 mg. The DDD facilitates a direct dose comparison between different thiazides.


The following covariables were assessed as potential confounders or effect modifiers: sex, age, BMI, estimated glomerular filtration rate (eGFR), heart failure, and systolic and diastolic blood pressure. Clinical measurements were obtained during the visits at the Rotterdam Study research center. BMI was expressed as kilograms per meter squared, and measurements (height and weight) were obtained during examination at the center visits. Assessment of heart failure in the Rotterdam Study has been described in detail previously.[20]

Serum creatinine concentration was obtained at the time of serum sodium measurement. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine equation[21,22] and was expressed in mL/min/1.73 m2. For the analyses in strata, the continuous variables BMI and eGFR were divided into quartiles. For age, we used predefined categories (≤55, >55–65, >65–75, >75–85, and >85 years). eGFR values were present in only a subset of participants with a serum sodium measurement; missing BMI data were random.

Statistical Analysis

The association between thiazides and hyponatremia was evaluated using Cox proportional hazard regression analyses with exposure to thiazides as time-varying determinant, matched on follow-up time since May 1997.[23] In this model, thiazide exposure in each case of hyponatremia is compared with all other participants within the cohort who did not have hyponatremia, with the same duration of follow-up (Fig 1). Risks were expressed as hazard ratios (HRs) plus 95% confidence intervals (CIs) and additionally were analyzed separately for mild, moderate, and severe hyponatremia. Cox proportional hazard models were adjusted for the confounding effect of sex and age, which were, in contrast to BMI and eGFR, both associated with a >10% change of the estimate. In an additional model, we adjusted for heart failure and systolic and diastolic blood pressure. These analyses were performed within separate strata of sex, age, DDD, BMI, and eGFR. Effect modification by sex, age, BMI, and eGFR (multiplicative interaction) was tested separately with interaction terms. Additionally, sensitivity analysis was performed, in which cases were compared with all participants with a sodium measurement between a week before and a week after the index date. We tested for proportionality by studying interaction terms with follow-up time. Statistical analyses were performed by SPSS software (version 17.0; SPSS Inc). P < 0.05 was considered statistically significant.

Figure 1.

Study design; risk of hyponatremia in those using thiazides versus those not using thiazides, assessed in a case and at the same time of follow-up in the rest of the cohort, after which the case is censored. Participants in the Rotterdam Study (RS) from the 3 cohorts were included (RSI, RSII, and RSIII). Thiazide exposure is illustrated by the dotted vertical lines and cases are illustrated by an arrow.