This retrospective study included inpatients with schizophrenia admitted between January 2015 and April 2022. Patients met the diagnosis criteria of schizophrenia according to the International Classification of Diseases-10 (ICD-10) and received mECT treatment during their hospitalization. The diagnosis of HAP required all the following criteria: new lung infiltrates on chest imaging, respiratory decline, fever, and productive cough (37). Patients with infections within 48 h of hospitalization were excluded. This study was approved by the Ethics Committee of the Fourth People's Hospital of Chengdu.

Patient information collected included name, age, gender, as well as the status of diabetes mellitus, hypertension, epilepsy, or substance dependence (smoking or drinking), and excluded patients with comorbid cardiovascular disease. Blood samples were collected on admission for routine biochemical testing (white blood cells, red blood cells, platelets, lipids, glucose, blood proteins, etc.). Other medications of patients receiving mECT at the time of hospitalization were recorded, such as sedative-hypnotic drugs (SHD), antidepressant drugs (ADD), anti-anxiety drugs (AAD), antimanic drugs (AMD), anti-epileptic drugs (AED), anti-parkinsonian drugs (APD), and other neurological drugs. mECT-related conditions only for patients with HAP, such as the days from the first mECT to HAP (dfE2H) occurrence, the numbers of mECT treatments before HAP (nE2H) occurrence, and the days from the last mECT treatment to HAP (dlE2H) occurrence.

The ECT Instrument was Thymatron System IV (Somatics, LLC, 149 Amityville Street Islip Terrace, NY 11752 USA). Electrode placement was a bilateral temporal energization mode. The electrical stimulation procedure was LOW 0.5: Frequency: 10–70 Hz; Pulse width: 0.5 ms; Duration: ≤ 8s; Waveform: bipolar, brief pulsed, square wave. Stimulus intensity (power): “Age mode” is used. For the first treatment session, the power is set to age × 80% for younger than 30 and age × 100% for older than 50; The power of follow-up treatment increased by 1–5% depending on the seizure index. Convulsions quality assessments: EEG Endpoint was 20–60 s; Average Seizure Energy Index was over 5,000; Postictal Suppression Index was over 80%.

Software SPSS 26 (IBM Corporation, New Orchard Road, Armonk, NY 10504, USA) was used for statistical calculations. These patients were divided into two groups by gender, with subgroups of HAP and non-HAP. Then the statistical analysis was conducted as follows: First, a general linear model (univariate model) was used for assessing the interaction effect of gender vs. other factors on HAP. Second, influencing factors for HAP men and women were analyzed separately. The χ² test was used for categorical variables. Continuous variables were first tested for normality by Kolmogorov-Smirnov; if variables conformed to a normal distribution, t-testing was used, while non-normality variables were tested by non-parametric (Mann-Whitney) test. Since the included data were almost all non-normal variables, Spearman correlation analysis was used. Third, binary logistic regression modeling was later used for risk factors analysis. Finally, statistical calibration was performed using the Bonferroni method (P < 0.05/31 ≈ 0.0016). Continuous variables were expressed as mean ± standard deviation (x̄ ± std.), and P < 0.05 was considered to be statistically significant.

A total of 951 inpatients, aged between 14 and 71 years old, were included. Of the 951 inpatients, 375 were male, and 576 were female, with mean ages of 32.57 ± 11.93 years and 36.85 ± 13.96 years, respectively, and males being significantly younger than females (t = 5.053, P < 0.001). 62 inpatients were HAP, with 37 in men and 25 in women. Covariance analysis results indicate interaction between sex and epilepsy, TC, anti-depressant drugs, anti-epileptic drugs, anti-parkinsonian drugs, and other neuron drugs (all Ps < 0.05), as shown in Table 1. Only epilepsy passed the Bonferroni test.

HAP occurred in 62 of the 951 study subjects, with an incidence of 6.52%. Patients developed HAP 20.10 ± 18.29 days after admission, received about 3.47 ± 2.57 sessions of mECT treatment before HAP, and HAP occurred approximately 2.78 ± 4.70 days after the last MECT treatment. The prevalence of HAP was significantly higher in men than in women (37/375:25/576 = 9.87%:4.34% ≈ 2.3, χ² = 11.382, P < 0.001). There was no statistically significant difference between men and women in dfE2H, nE2H, or dlE2H, as shown in Table 2.

In Table 3, hospitalized days were significantly higher in the HAP group, as compared to the non-HAP group (Z = −2.760, P = 0.006). Total cholesterol (mmol/L) was significantly lower in the HAP group as compared to the non-HAP group (Z = −2.147, P = 0.032), while monocyte count was significantly higher (Z = −2.001, P = 0.045). There was a statistically significant difference in basophils between the HAP and non-HAP group (Z = −2.065, P = 0.039). There were also statistically significant increases in HAP incidence found in patients who had taken any medications of SHD (χ² = 11.396, P < 0.001), ADD (χ² = 5.237, P = 0.002), AED (χ² = 7.993, P = 0.005), APD (χ² = 17.973, P < 0.001), or other neurological drugs (χ² = 7.323, P = 0.007), though only SHD and APD passed the Bonferroni test.

There were correlations between HAP and days in hospital (r_s = 0.143, P = 0.006), total cholesterol (r_s = −0.111, P = 0.032), monocyte count (r_s = 0.103, P = 0.045), SHD (r_s = 0.174, P = 0.001), ADD (r_s = 0.118, P = 0.022), AED (r_s = 0.146, P = 0.005), APD (r_s = 0.219, P < 0.001), and other neurological drugs (r_s = 0.140, P = 0.007) by Spearman test. After input of these variables into the logistic regression equation, only total cholesterol [Beta = −0.373, Wald = 3.920, P = 0.048, Exp(B) = 0.688 (0.476, 0.996)] and APD [Beta = 1.366, Wald = 14.84, P < 0.001, Exp(B) = 3.920 (1.925, 7.981)] survived.

In Table 4, statistically significant higher incidences of diabetes (χ² = 6.203, P = 0.013), hypertension (χ² = 9.096, P = 0.003), epilepsy (χ² = 5.643, P < 0.001), lymphocyte count (Z = −2.408, P = 0.016), eosinophil count (Z = −2.141, P = 0.032), SHD (χ² = 13.636, P < 0.001), and APD (χ² = 8.526, P = 0.004) were identified in the HAP group, as compared to the non-HAP group, while lower levels of total bilirubin (Z = −2.024, P = 0.043) on admission were found in the HAP group. Only variables of epilepsy and SHD passed the Bonferroni test.

Spearman testing showed that HAP is correlated with diabetes (r_s = 0.104, P = 0.013), hypertension (r_s = 0.126, P = 0.003), epilepsy (r_s = 0.196, P < 0.001), total bilirubin (r_s = −0.084, P = 0.043), lymphocyte count (r_s = −0.100, P = 0.016), eosinophil count (r_s = −0.089, P = 0.032), SHD (r_s = 0.154, P < 0.001), and APD (r_s = 0.122, P = 0.003). After input of these variables into the logistic regression equation, only lymphocyte count [Beta = 1.702, Wald = 5.432, P = 0.020, Exp(B) = 5.483 (1.311, 22.937)], hypertension [Beta = −0.835, Wald = 4.764, P = 0.029, Exp(B) = 0.434 (0.205, 0.918)], and SHD [Beta = 2.287, Wald = 9.387, P = 0.002, Exp(B) = 9.847 (2.280, 42.536)] survived.

Our results showed that among HAP patients who underwent mECT, there was no statistical difference between men and women in the three indicators: days from first mECT treatment to HAP occurrence (dfE2H), the numbers of mECT treatments before HAP occurrence (nE2H), and days from last mECT treatment to HAP occurrence (dlE2H). However, the results showed that the median number of days from the last mECT treatment to HAP was 1 day in both men and women, indicating that all patients had a very high risk of developing HAP within 1 day after receiving mECT treatment. In addition, the risk of HAP within the first three mECT treatments was quite high in men and women, respectively, which may be due to the fact that patients need to gradually adapt to the clinical symptoms that may occur after mECT and ignore the possible risk of HAP. Therefore, healthcare providers should be on alert the first day after each mECT treatment, and special attention should be paid to clinical care after the first 3 sessions of mECT treatment.

Interestingly, we found that male patients with SCZ are more prone to HAP than women, and the risk factors predisposing both groups to HAP also differ from each other.

The prevalence of HAP in male mECT patients in this study was significantly higher than that in women (male incidence ~2.3 times that of females). This increased prevalence may be related to lifestyle habits of male patients, such as smoking, alcohol abuse, low weight, frequent contact with children, and poor oral hygiene, which are all risk factors for HAP (47). In addition, this study found that lower total cholesterol prior to hospital admission and APD medication in hospitalization may be risk factors for the development of HAP in men.

Our results showed that male SCZ patients treated with mECT with lower total cholesterol levels at admission were more likely to develop HAP, and logistic regression analysis showed that low levels of total cholesterol might be an independent risk factor for HAP in SCZ patients treated with mECT. Cholesterol is widely distributed in many tissues of the body, especially the brain and neural tissues, and may be associated with a variety of diseases (48), and even mental status or personality changes (49). Importantly, cholesterol has an important role in coronavirus entry, membrane fusion, and pathological syncytium formation. 25-hydroxycholesterol (25HC) is one of the metabolites of cholesterol, and 25HC inhibits coronavirus infection by blocking membrane fusion (50), so lower total cholesterol levels may lead to lower 25HC levels, increasing the chance of coronavirus infection in SCZ patients and potentially explaining the correlation between viral infection and lower cholesterol levels (51, 52). Lower total cholesterol may also be a factor of increased short-term mortality in elderly patients with community-acquired pneumonia (CAP) (53). In addition, men may be more sensitive to low levels of cholesterol (54), which accounts for the fact that men with low total cholesterol levels were more likely to develop HAP in our study, whereas women showed no statistical difference in HAP based on the level of total cholesterol.

APDs are used to improve Parkinsonian-like symptoms and extrapyramidal effects in schizophrenic patients. Our results show that male mECT SCZ patients using APDs during hospitalization were more likely to develop HAP, and logistic regression analysis showed that APD use is an independent risk factor for the incidence of HAP. Relatedly, it has been shown that anticholinergic drugs may increase the risk of HAP (55); despite some studies showing that amantadine can be used to prevent pneumonia (56), the presence of drowsiness, falls, and skin problems (57) may in turn lead to an increased risk of developing HAP in patients.

Risk factors for HAP in women are hypertension, low lymphocyte count on admission, and use of SHDs during hospitalization. There was no statistically significant difference in the prevalence of hypertension between men and women in patients treated with mECT (9/375:15/576, χ² = 0.038, P = 0.844), however, SCZ patients with comorbid hypertension in women had a higher risk of HAP, compared to men. ECT treatment may not result in significant changes in blood pressure in hypertensive patients (58), however, patients with comorbidities such as chronic renal insufficiency or diabetes mellitus are more likely to be affected by COVID-19, potentially fatally (59–61). Therefore, for female SCZ patients with comorbid diseases such as hypertension, clinical management should be particularly strengthened during hospitalization to reduce the risk of HAP and improve the quality of patient survival.

Lymphocytes have an important role in immune regulation and can be involved in the pathogenesis of respiratory diseases such as pneumonia, infections, asthma, and acute respiratory distress syndrome (62). The lower lymphocyte count in female mECT patients suggests that immunity may be reduced, potentially increasing the incidence of HAP.

Benzodiazepines drugs (BZD) are often used in sedation-hypnosis and can modulate peripheral γ-aminobutyric acid (GABA) type A on macrophages and increase the incidence of infection by inhibiting proinflammatory cytokines (63). Experiments performed on mice have further shown that benzodiazepines enhance GABA signaling, leading to increased mortality from pneumonia (64). In addition, certain meta-analyses have shown an increased risk of pneumonia with recent or current exposure to benzodiazepines (65, 66), which is consistent with our findings. Therefore, BZDs should be used with caution in SCZ patients receiving mECT.

Although mECT is used as a routine treatment for patients with schizophrenia, the number of patients receiving it remains low, which may stem from the fear of electrical stimulation. Since the outbreak of COVID-19, more standardized clinical management and increased awareness of personal protection implemented by medical institutions have led to a decrease in the number of patients with mECT experiencing HAP; therefore, random errors induced by small sample size may have a greater impact on the statistical analysis. Incomplete demographic indicators for some patients, such as height, weight, and education, as well as the lack of cognitive assessment of patients with schizophrenia during data collection, prevented our results from fully reflecting the full picture of patients. Data from an individual psychiatric hospital is another limitation, as regional differences, local culture, economic conditions, and ethnic groups of patients may also influence our results. Therefore, these risk factors deserve to be investigated as a prospective study with a larger patient sample size and collaboration of multiple clinical centers.