When the Water Stops: Dialysis Service Disruption During Catastrophic Urban Flooding in Southern Thailand

Introduction

In late November 2025, Hat Yai, the largest metropolitan center in southern Thailand, experienced catastrophic urban flooding following extreme monsoonal rainfall and cloudburst events that far exceeded historical records.S1–S3 Rapid inundation overwhelmed drainage systems, disrupted transportation, and triggered near-simultaneous failure of electricity, municipal water supply, telecommunications, and health care infrastructure across Songkhla Province. Beyond acute injuries and infectious risks, the disaster severely disrupted continuity of chronic disease care, with substantial loss of life and mass displacement reported across the region.S4^,S5

Patients with kidney failure are among the most disaster-vulnerable populations, because kidney replacement therapy cannot be safely interrupted. In Thailand, > 120,000 patients receive long-term dialysis under the Universal Coverage Scheme.^1,2 Thailand’s “PD-first policy” designates peritoneal dialysis (PD) as the initial modality, supported by centralized logistics and home-based care delivery, reducing dependence on facility-based infrastructure.³ However, dialysis delivery remains highly dependent on infrastructure, supply chains, and workforce capacity, rendering it vulnerable to climate-related disruptions.⁴

Despite global emphasis on dialysis disaster preparedness, real-world system-level evaluations remain limited, particularly in middle-income settings.5, 6, 7 We conducted a province-wide postdisaster survey of all dialysis centers in Songkhla to quantify service disruption, characterize system vulnerabilities, and derive actionable preparedness strategies. The detailed methods are provided in the Supplementary Material.

Results

All 20 dialysis centers participated (100% response rate), including public, university, military, and private facilities. Baseline characteristics are summarized in Supplementary Table S1. Before the flooding, the province operated 287 functional hemodialysis (HD) machines and provided PD care to 435 patients. During the peak flood period (November 22–December 5, 2025), 11 of 20 centers (55%), all located in Hat Yai, ceased operations, demonstrating marked geographic clustering of failure (Figure 1). Median service interruption was 4 days (interquartile range: 2–9).

Figure 1.

Flood-related cascade of dialysis service disruption during the November 2025 floods in Songkhla Province.

Service suspension was driven primarily by loss of municipal water supply, physical inaccessibility, and cascading infrastructure failure, including electricity and telecommunications disruption (Supplementary Table S2 and Figure 2). Structural damage occurred exclusively in nonoperational centers. All operational centers functioned under crisis standards of care, defined as adaptive modifications to dialysis delivery (e.g., reduced duration, frequency, or dialysate flow) under resource constraints to preserve capacity and prioritize urgent care. No center maintained standard HD practice (Supplementary Table S3). Two mobile HD units, deployed from Bangkok within 48 hours, each operated 2 machines and delivered approximately 8 treatments/d, totaling approximately 50 to 60 sessions over 7 days, partially mitigating capacity loss. However, system strain was substantial, with some centers delivering up to 200 HD sessions daily under prolonged staff working hours.

Figure 2.

(a) Primary barriers to hemodialysis service continuity and (b) infrastructure failures by center operational status during the November 2025 floods. Center-level data show clustering of access and infrastructure failures among nonoperational centers, with physical inaccessibility and loss of municipal tap water supply as the predominant barriers during the acute flood period. RO, reverse osmosis.

Overall, 347 HD patients (39%) and 91 PD patients (21%) were affected; these figures represent center-reported aggregate estimates rather than individually verified patient-level data and should be interpreted with caution given potential under- and over-ascertainment. Patients in nonoperational centers experienced higher rates of missed dialysis and complications, including pulmonary edema, hyperkalemia, and uncontrolled hypertension (Supplementary Table S4).

Mortality analysis demonstrated numerically higher deaths in nonoperational centers, although not statistically significant (HD mortality, 1.7% [3/180] vs. 0.8% [1/128] [P = 0.6]; PD mortality, 1.1% [2/189] vs. 0% [0/198] [P = 0.2], in nonoperational and operational centers, respectively). These findings are exploratory and should not be interpreted as causal, given the limited statistical power and absence of baseline mortality comparators. PD services demonstrated relative continuity but remained vulnerable to supply disruption, equipment damage, and unsafe environments for sterile exchange. These observations are descriptive and should not be interpreted as evidence of modality superiority (Supplementary Table S4).

Workforce capacity was severely strained. Standardized staff-level denominators were not available; therefore, center operational status was used as a proxy for workforce availability. Service disruption was primarily infrastructure-driven, though operational centers maintained services through extended hours and shift intensification. Staff reported prolonged working hours with limited rest, supported by exploratory, nonvalidated center-level assessments (Supplementary Table S5).

A district-level “War Room” incident command system enabled centralized triage, interfacility redistribution, and mobile dialysis deployment, supported by national coordination networks. However, telecommunication failure emerged as the principal bottleneck limiting system responsiveness.

Discussion

This province-wide assessment demonstrates that dialysis disasters are fundamentally infrastructure and systems failures rather than deficits in clinical expertise. Loss of municipal water supply emerged as the most critical single point of failure, immediately disabling HD services even when other infrastructure remained partially intact.⁵ Although Thailand’s PD-first policy reduced reliance on facility-based HD, it introduced alternative dependencies, including supply chain continuity and digital infrastructure. PD resilience was therefore conditional, with vulnerabilities exposed under system-wide disruption.^6,7

An important finding was the near-universal impact on the health care workforce, whose members sustained service delivery despite personal loss and extreme working conditions, underscoring the need to formally integrate workforce protection into disaster preparedness frameworks.^3,5

Postdisaster, the evaluation identified both strengths and gaps. Rapid frontline adaptation, informal interfacility collaboration, and mobile dialysis deployment were key strengths. However, the absence of predefined command structures, delayed coordination, and concurrent infrastructure collapse limited early response effectiveness. These findings inform a structured “Hat Yai Model” for dialysis disaster preparedness, comprising the following: (i) data: real-time dialysis registry; (ii) command: predefined incident command (“War Room”); (iii) communication: redundant communication systems; (iv) resources: mobile dialysis capacity and workforce protection; and (v) recovery: infrastructure redundancy, particularly water and power systems.

All statistical comparisons are exploratory and unadjusted for multiple testing, and findings should be interpreted as hypothesis-generating. Limitations of the study include reliance on center-reported data, potential reporting bias, lack of patient-level verification, inability to determine causality of deaths, absence of standardized workforce denominators, and restriction to a single province, limiting generalizability.

As climate-related disasters intensify globally, particularly in urban centers of middle-income countries, failure to invest in integrated dialysis preparedness will translate into preventable morbidity and mortality among dialysis-dependent populations.

Disclosure

TK has received consultancy fees from Visterra as a country investigator and is currently supported by funding from the National Research Council of Thailand. He has received speaking honoraria from AstraZeneca, Baxter Healthcare, and Fresenius Medical Care. All the other authors declared no competing interests.

Supplementary Material

Supplementary File (PDF)

Supplementary Methods. Supplementary References. Table S1. Summary of baseline facility characteristics of participating hemodialysis units, stratified by operational status during the flood period (operational vs. nonoperational). Table S2. Barriers to dialysis service continuity and infrastructure/system failures during the November 2025 floods, stratified by modality and center operational status. Table S3. Adaptive strategies for dialysis service delivery during the November 2025 floods, stratified by modality and facility operational status. Table S4. Adverse impacts among dialysis patients during the November 2025 floods, stratified by modality and center operational status. Table S5. Center-reported workforce strain domains during the November 2025 floods, stratified by dialysis center operational status. Dialysis Service Disruption and Disaster Response Survey (DSD-DRS).STROBE Checklist.