Induced Seismicity and Acute Development of Anxiety and Stress-Related Mental Disorders: Findings from the 2017 Pohang Earthquake.

Introduction

An earthquake with a moment magnitude () of 5.4 occurred in Pohang, Korea, on 15 November 2017, and over 600 aftershocks occurred.[1] This series of earthquakes was the most damaging disaster Korea has experienced, causing a property loss of over $55 million USD.[2] The Korean government found that the earthquake was a result of induced seismicity triggered by the geothermal plant in Northern Pohang.[3]

Because of the sudden and threatening nature of earthquakes, residents may develop anxiety-related disorders.[4] Previous disaster studies have attempted to estimate the effects of earthquakes on the development of mental disorders; however, they have generally been limited by post-event designs and lack of proper controls.[5] For example, one study addressing the mental health effects of the Pohang earthquake interviewed 200 Pohang residents 18 months after the earthquake and observed an association between posttraumatic stress and degree of earthquake exposure.[1] A few studies have been conducted with unexposed controls; however, these studies did not have preearthquake data.[6,7]

The health insurance system of Korea covers the entire population and allows the construction of a retrospective cohort. The use of the residential address and hospital use information of each individual enabled the definition of the earthquake-exposed and nonexposed groups and the exact timing of the disease development. The association between the 2017 Pohang earthquake and the acute development of anxiety and stress-related disorders was evaluated using the National Health Insurance Database of Korea. To confirm the association, the analysis was replicated using the 2016 Kyungju earthquake, the nearest independent earthquake.

Methods

Design and Data

The difference-in-difference framework and the National Health Insurance Database was used. One city (Gimhae) located far away, which has a similar population size and similar medical infrastructures as Pohang, and one city (Kyungju) closest to Pohang, were selected as controls (Figure 1). The increase in the disorder development in Pohang residents was expected after the earthquake. In addition, the residents living closer to the epicenter (Northern Pohang) may have been more affected.

Figure 1.

(A) shows the map of the study area (Pohang, Kyungju, and Gimhae) and distribution of damage by the Pohang and Kyungju earthquakes (dots). (B) shows the daily incidence patterns of neurotic, stress-related, and somatoform disorders prior to ( to ) and after (0–80 d) the Pohang earthquake in Pohang, Kyungju, and Gimhae. The dashed vertical line represents the day the Pohang earthquake occurred. (C) shows the weekly incidence patterns of neurotic, stress-related, and somatoform disorders prior to ( to ) and after (0–100 wk) the Pohang earthquake in Pohang and Kyungju. Dashed vertical lines represent the weeks the Kyungju and Pohang earthquakes occurred.

The earthquake with the highest of nation was the Kyungju earthquake (: 5.8) occurred on 12 September 2016. Unlike the Pohang earthquake, the damage of this earthquake was distributed at a broad range (Figure 1). Therefore, an increase in the disorder development among Kyungju and Pohang residents was expected after the Kyungju earthquake.

Medical information spanning 10 y (2010–2019) for 50% of randomly selected residents who resided in Pohang and the control cities in 2014 was acquired. A total of 515,750 residents who did not move until death or the end of 2019 were evaluated. Past medical history from each individual was screened to estimate new disease development after the earthquake. The hospital visits with diagnosis for neurotic, stress-related, and somatoform disorders were selected based on the International Statistical Classification of Diseases and Related Health Problems, 10th Revision[8] (ICD-10 codes F40.X–F48.X). This study was exempted from review by the institutional review board of the Chungnam National University (202106-SB-084-01).

Statistical Analysis

The study period was divided into 42 d prior to and after the initial Pohang earthquake. The following model was used:

The daily number of the incident patients on day () was assumed to follow a quasi-Poisson distribution. is an indicator variable for the region (: Pohang and : control city), is the natural spline term applied to the time elapsed since the start of the study period, and log (pop) is an offset term for the logarithm of the number of population with health insurance qualification. is an indicator variable for the study period prior to and after the earthquake ( to , –20 d, and –41 d), and is an interaction term between region and period indicators. to were exponentiated to estimate difference-in-difference estimates. Stratification analysis by sex and age (0–19, 20–59, and y) was performed. The level of statistical significance was set at a -value below 0.05.

Results

A total of 1,876 anxiety and stress-related disorder events were identified prior to ( to ) and after (0–41 d) the Pohang earthquake [Pohang: 782 incidences (Northern Pohang: 490; Southern Pohang: 292); Kyungju: 388 incidences; Gimhae: 706 incidences]. An increase in the daily number of the disorder incidences was observed in Pohang and Northern Pohang within 0–20 d after the initial Pohang earthquake (Figure 1). Compared with Gimhae and the time period prior to the earthquake ( to ), a 2.08-fold [95% confidence interval (CI): 1.19, 3.64] increase in the relative risk (RR) for the disorder development in Pohang was observed within 0–20 d after the earthquake (Table 1). About 52% (170 of 328) of newly diagnosed anxiety and stress-related disorder patients within the 0–20 d after the earthquake in Pohang were estimated to be attributable to the earthquake. Compared with Kyungju, a similar increase in risk was observed [ (95% CI: 1.22, 4.40)].

Table 1

Association between the Pohang earthquake and changes in the daily incidence patients of neurotic, stress-related, and somatoform disorders prior to ( to ; to ) and after (0–20 d; 21–41 d) the earthquake (difference-in-difference comparison between Pohang, Kyungju, and Gimhae).

Time period (d)	Daily average incidence rate (/100,000 persons)					Difference-in-difference estimatesa [RR (95% CI)]
	Pohang	Northern Pohang	Southern Pohang	Gimhae	Kyungju	Pohang vs. Gimhae	Pohang vs. Kyungju	Northern Pohang vs. Kyungju	Southern Pohang vs. Kyungju	Gimhae vs. Kyungju
Total
−42 to −22	3.68	3.65	3.73	3.99	4.84	—	—	—	—	—
−21 to −1	3.39	4.01	2.68	3.95	4.20	0.93 (0.50, 1.73)	1.06 (0.53, 2.12)	1.27 (0.61, 2.63)	0.83 (0.41, 1.66)	1.14 (0.63, 2.07)
0–20	8.06	10.52	5.25	4.20	4.57	2.08 (1.19, 3.64)	2.32 (1.22, 4.40)	3.06 (1.58, 5.92)	1.49 (0.80, 2.79)	1.11 (0.62, 2.00)
21–41	4.08	4.43	3.68	4.06	4.11	1.09 (0.60, 1.99)	1.30 (0.66, 2.59)	1.43 (0.70, 2.95)	1.16 (0.60, 2.25)	1.20 (0.66, 2.18)
Men
−42 to −22	3.01	2.64	3.42	3.58	4.26	—	—	—	—	—
−21 to −1	2.67	2.73	2.59	3.76	2.69	0.84 (0.43, 1.63)	1.40 (0.63, 3.15)	1.64 (0.68, 3.94)	1.20 (0.50, 2.88)	1.67 (0.81, 3.44)
0–20	4.35	5.18	3.42	3.26	3.43	1.58 (0.84, 2.99)	1.79 (0.85, 3.76)	2.44 (1.11, 5.37)	1.24 (0.55, 2.80)	1.13 (0.56, 2.28)
21–41	2.77	3.30	2.18	3.26	4.54	1.01 (0.52, 1.97)	0.86 (0.41, 1.81)	1.17 (0.53, 2.62)	0.60 (0.25, 1.40)	0.85 (0.44, 1.67)
Women
−42 to −22	4.35	4.61	4.04	4.40	5.40	—	—	—	—	—
−21 to −1	4.11	5.25	2.76	4.13	5.67	1.01 (0.49, 2.07)	0.90 (0.41, 1.95)	1.08 (0.48, 2.44)	0.65 (0.29, 1.47)	0.89 (0.45, 1.77)
0–20	11.73	15.64	7.13	5.14	5.67	2.31 (1.22, 4.37)	2.57 (1.25, 5.27)	3.23 (1.54, 6.78)	1.68 (0.83, 3.41)	1.11 (0.57, 2.17)
21–41	5.38	5.52	5.21	4.86	3.69	1.12 (0.56, 2.22)	1.81 (0.80, 4.08)	1.75 (0.74, 4.11)	1.89 (0.87, 4.09)	1.62 (0.78, 3.34)
Age 0–19 y
−42 to −22	1.82	1.47	2.23	2.57	2.85	—	—	—	—	—
−21 to −1	1.36	0.63	2.23	2.32	2.37	0.83 (0.28, 2.46)	0.90 (0.28, 2.90)	0.51 (0.10, 2.58)	1.20 (0.36, 4.05)	1.08 (0.36, 3.26)
0–20	3.18	3.36	2.97	2.40	2.13	1.87 (0.72, 4.86)	2.33 (0.77, 7.04)	3.05 (0.87, 10.71)	1.78 (0.54, 5.84)	1.24 (0.40, 3.84)
21–41	0.79	0.42	1.24	1.89	1.42	0.60 (0.17, 2.07)	0.88 (0.22, 3.47)	0.57 (0.09, 3.74)	1.11 (0.27, 4.64)	1.47 (0.42, 5.17)
Age 20–59 y
−42 to −22	3.61	3.41	3.84	4.15	4.40	—	—	—	—	—
−21 to −1	3.57	4.60	2.41	4.64	3.98	0.88 (0.45, 1.73)	1.09 (0.50, 2.37)	1.49 (0.66, 3.34)	0.69 (0.32, 1.51)	1.23 (0.64, 2.40)
0–20	8.57	11.43	5.36	4.91	4.90	2.01 (1.09, 3.69)	2.13 (1.05, 4.32)	3.01 (1.45, 6.25)	1.25 (0.64, 2.47)	1.06 (0.56, 2.01)
21–41	4.04	3.89	4.20	4.57	4.40	1.01 (0.52, 1.97)	1.12 (0.52, 2.38)	1.14 (0.51, 2.57)	1.09 (0.54, 2.22)	1.10 (0.57, 2.11)
Age ≥60 y
−42 to −22	5.92	6.73	5.00	6.23	7.28	—	—	—	—	—
−21 to −1	5.05	6.03	3.95	4.03	6.03	1.32 (0.56, 3.11)	1.03 (0.44, 2.42)	1.08 (0.42, 2.77)	0.95 (0.35, 2.59)	0.78 (0.33, 1.85)
0–20	11.84	15.78	7.37	4.58	5.68	2.72 (1.24, 5.96)	2.56 (1.15, 5.70)	3.00 (1.28, 7.07)	1.89 (0.76, 4.71)	0.94 (0.40, 2.20)
21–41	7.77	10.44	4.74	6.23	5.50	1.31 (0.61, 2.82)	1.74 (0.76, 3.98)	2.05 (0.84, 5.00)	1.25 (0.47, 3.33)	1.32 (0.58, 2.99)

Note: —, not applicable; CI, confidence interval; RR, relative risk.

Difference-in-difference model adjusting for time trend and population size.

In a subregional analysis, a 3.06-fold (95% CI: 1.58, 5.92) increase in the RR for disorder development in Northern Pohang was observed after the earthquake when compared with Kyungju. The increase in RR was prominent in women; however, no increase was observed when Southern Pohang and Kyungju, and Kyungju and Gimhae were compared.

Weekly anxiety and stress-related disorder incidence pattern showed an increase for Kyungju and Pohang residents after the Kyungju earthquake; however, the increase was limited in Pohang after the Pohang earthquake (Figure 1). Corresponding to the wide geographical distribution of the damage caused by the Kyungju earthquake, an increase in disorders was observed in both Pohang and Kyungju residents after the Kyungju earthquake. The increase in the disease incidence lasted for a short period (within 3 wk) after the Kyungju and Pohang earthquakes.

Discussion

The 2017 Pohang earthquake was an induced seismicity triggered by the activity of a geothermal plant in Northern Pohang. By using a retrospective cohort based on the health insurance data, a 2-fold increase in the risk of anxiety and stress-related disorder development was observed days after the Pohang earthquake. Subregional analyses showed an increase in Northern Pohang residents but not in residents of Southern Pohang and other control cities, suggesting a clear exposure–response relationship.

To our knowledge, none of the previous earthquake studies replicated their findings with subsequent earthquakes using the same data set. Similar to the results of the Pohang earthquake, an increase in the risk of anxiety and stress-related disorder development was observed 0–20 d after the Kyungju earthquake.

The existence of the time-varying factors coinciding with the earthquake may limit the causal interpretation of the results. Although the time-varying factors that influenced both Pohang and the control cities can be canceled out by the difference-in-difference study design, there could be factors specific to Pohang.