Second-generation antipsychotics, sometimes called atypical antipsychotics, were introduced in part because they promised a more favorable side-effect profile than the older first-generation drugs, which were notorious for causing movement disorders. Previous research had hinted that the second-generation class as a whole might carry a lower risk of hyponatremia than the first-generation agents. What remained unclear, however, was whether the risk differed meaningfully between individual drugs within the class. This distinction matters enormously in clinical practice. If all second-generation antipsychotics carried the same hyponatremia risk, clinicians could choose among them based on other considerations. But if the risk varies from one molecule to the next, that variation could become a decisive factor when prescribing for patients who are already vulnerable to sodium disturbances, such as older adults, people taking other medications that affect sodium levels, or those with underlying medical conditions.

Addressing this question, a research team led by Associate Professor Masakazu Hatano of the Department of Pharmacotherapeutics and Informatics at Fujita Health University School of Medicine in Japan, together with Dr. Takenao Koseki, Dr. Takeo Saito, and Dr. Shigeki Yamada, all affiliated with the same institution, carried out a two-stage comparative safety study. Their strategy combined two very different types of data, each with distinct strengths and weaknesses, to arrive at conclusions more robust than either source could provide alone. Hyponatremia is a recognized safety concern associated with antipsychotic use, but as Dr. Hatano explained, comparative evidence among individual second-generation antipsychotics has remained limited. The team therefore adopted a two-stage approach, first using a spontaneous reporting system to generate hypotheses about hyponatremia risk among individual drugs, and then using a claims database to evaluate the associations suggested by the first analysis.

The hypothesis-generating stage drew on the US Food and Drug Administration Adverse Event Reporting System, a vast pharmacovigilance repository that collects reports of suspected side effects from manufacturers, healthcare professionals, and patients. The researchers examined reports spanning from the fourth quarter of 1997 to the third quarter of 2023, an enormous window covering more than a quarter century of clinical use. In total, the analysis included 17,805,418 reports, of which 458,745 involved second-generation antipsychotic monotherapy, meaning patients taking a single antipsychotic rather than a combination. Using a statistical technique known as disproportionality analysis, which compares how often a given adverse event is reported for one drug relative to others, the team calculated reporting odds ratios for hyponatremia with olanzapine as the reference drug. Several antipsychotics showed lower reporting odds than olanzapine, and brexpiprazole and lurasidone showed the lowest reporting odds of all.

However, the researchers were careful to emphasize that these findings could only be considered hypothesis-generating rather than definitive. Spontaneous reporting systems suffer from a fundamental limitation: they provide no denominator information. In other words, while the database records how many adverse event reports mention a particular drug, it does not record how many people in total took that drug. Without knowing the size of the exposed population, it is impossible to calculate actual incidence or absolute risk. Reporting is also vulnerable to biases, including heightened attention to certain drugs, differences in how adverse events are recognized and attributed, and the passage of time as drugs gain or lose market share. A drug that is newly marketed or heavily scrutinized may generate more reports simply because clinicians are watching it more closely, not because it is genuinely more dangerous.

This is precisely why the second stage of the study was essential. The researchers turned to a Japanese hospital-based claims database covering April 2008 to April 2024, which allowed them to observe real patients over time rather than relying on voluntary reports. From this database, they identified 55,394 patients who were newly prescribed antipsychotics, had no recent antipsychotic use, and had no existing hyponatremia, ensuring that the analysis captured new-onset events rather than pre-existing conditions. Each patient was followed for up to 180 days after starting treatment, a window long enough to capture events plausibly related to the medication while limiting the influence of unrelated changes over time. During the follow-up period, hyponatremia occurred in 366 patients, representing 0.7 percent of the cohort, a figure that provides the kind of absolute risk context that spontaneous reporting data can never supply.

The critical analytical challenge in observational studies of this kind is confounding. Patients are not randomly assigned to antipsychotics; clinicians choose drugs based on patient characteristics, illness severity, prior treatment history, and anticipated tolerability. These same characteristics may independently influence hyponatremia risk. To address this, the researchers used statistical weighting methods to account for differences in baseline characteristics and other factors influencing risk, effectively creating balanced comparison groups. After adjustment, one finding stood out clearly: aripiprazole was associated with a significantly lower risk of hyponatremia than olanzapine. The other antipsychotics examined did not show significant differences compared with olanzapine once adjustment was applied, a notable contrast with the unadjusted signals from the reporting database, where several drugs had appeared more favorable.

These findings suggest that the risk of hyponatremia may differ among individual second-generation antipsychotics, Dr. Hatano clarified, adding that aripiprazole was associated with a lower risk than olanzapine in the adjusted analysis, which may help inform antipsychotic selection for patients at higher risk of hyponatremia. The result is pharmacologically plausible. Aripiprazole is a partial dopamine agonist with a distinctive receptor profile that sets it apart from most other second-generation agents, and olanzapine is known for broader receptor binding and a heavier metabolic and sedative burden. While the study was not designed to establish mechanism, the divergence between these two widely used drugs underscores that the second-generation label conceals real pharmacological heterogeneity with potential clinical consequences.

The researchers were candid about the limitations of their work. The study was observational, meaning it could identify associations but not prove causation. Residual confounding, the possibility that unmeasured differences between patient groups influenced the results, cannot be excluded, and differences in how hyponatremia was detected across treatment groups may also have shaped the findings. Laboratory monitoring practices, for example, may differ depending on the drug prescribed and the patient’s perceived risk, which could bias event ascertainment. The Japanese setting of the claims analysis raises questions about generalizability to other populations, and the 180-day follow-up window may not capture risks emerging over longer treatment periods. The authors also disclosed financial relationships with numerous pharmaceutical companies, which are documented in the published conflict of interest statement.

Despite these caveats, the study represents a valuable methodological template for drug safety research. By combining the breadth of a pharmacovigilance database with the clinical realism of claims data, the researchers generated and then tested hypotheses in a way that neither source could accomplish alone, providing complementary evidence for comparing drug safety while transparently acknowledging the limitations of each approach. For clinicians, the practical message is that hyponatremia risk deserves consideration when selecting among second-generation antipsychotics, particularly for patients with additional risk factors such as advanced age, thiazide diuretic use, or pre-existing sodium disturbances. For researchers, the findings highlight the need for further studies to elucidate the associations between individual second-generation antipsychotics and hyponatremia, ideally with designs capable of confirming mechanism and quantifying absolute risk across diverse populations. As antipsychotic prescribing continues to expand worldwide, studies of this kind bring precision to a class of drugs too often treated as interchangeable, moving psychiatric pharmacotherapy closer to genuinely individualized care.

Subject of Research: Comparative risk of hyponatremia among second-generation antipsychotics

Article Title: Hyponatremia risk may differ across second-generation antipsychotics

Article References: Hyponatremia risk may differ across second-generation antipsychotics. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: hyponatremia, second-generation antipsychotics, aripiprazole, olanzapine, pharmacovigilance, FAERS, claims database, drug safety, schizophrenia, bipolar disorder, JAMA Network Open, observational study