Epidemiology and Mortality-Related Prognostic Factors in Endophthalmitis

Abstract

Purpose: Endophthalmitis describes any intraocular inflammation that involves both the posterior and anterior segments and is divided into endogenous and exogenous types according to its pathogenesis. The incidence of endophthalmitis and its risk factors have been extensively evaluated. However, few studies have explored the mortality rate in patients diagnosed with endophthalmitis.

Methods: We obtained data entered into the National Health Insurance Research Database (NHIRD) from 2000 to 2013. The data collected included all discharge diagnoses of endophthalmitis in inpatients. Baseline characteristics, comorbidities, and prognostic factors were evaluated.

Results: This study identified 7764 patients who were diagnosed with endophthalmitis in Taiwan from 2000 to 2013. The mortality rate was 0.97% (75/7764), and the mean age was 63.57 ± 15.72 years. Epidemiological characteristics were compared as “with or without” for different systemic comorbidities, and the results indicated that the adjusted odds ratio (AOR) was significantly higher in cases comorbid with renal disease (AOR 2.864, P = 0.001), septicemia (AOR 8.886, P < 0.001), pneumonia (AOR 2.072, P = 0.030), and tumors (AOR 7.437, P < 0.001). However, comorbidity with diabetes mellitus (DM) lowered the AOR by 0.500-fold (P = 0.026). There was no significant difference in ORs between patients comorbid with hypertension, depression, anxiety, hyperlipidemia, thyrotoxicosis, liver disease, or injury (all P > 0.05).

Conclusions: Among inpatients with endophthalmitis, predictors of mortality include renal disease, septicemia, pneumonia, neoplasia, a greater burden of comorbidity (especially catastrophic illness), longer hospital stays (more than 11 days), and higher medical costs. Interestingly, DM decreased the OR for inpatient mortality.

The literal meaning of the term endophthalmitis is any intraocular inflammation that involves both the posterior and anterior segments. In clinical practice, endophthalmitis refers to an intraocular infection caused by microorganisms. Endophthalmitis is divided into endogenous and exogenous subtypes according to its pathogenesis. Endogenous endophthalmitis is caused by the hematogenous dissemination of bacterial, fungal, or other pathogens that break through the blood-ocular barrier and inoculate the intraocular region.^1–4 Patients with a compromised immune system are at risk of endogenous endophthalmitis. Predisposing conditions include diabetes mellitus (DM), systemic malignancy (especially lymphoma/leukemia), endocarditis, sickle cell anemia, autoimmune disease, and human immunodeficiency virus infection/acquired immunodeficiency syndrome (HIV/AIDS). Systemic immunomodulatory therapy, chemotherapy, longer hospital stays, and/or intensive care unit (ICU) admission have also been reported as risk factors. Possible sources of infection that should be considered are pneumonia, urinary tract infection, bacterial meningitis, and a liver abscess. In Asia, infection of a liver abscess with Klebsiella species is the most common cause, accounting for 60% of all cases of endogenous endophthalmitis.^2,4–6 

Exogenous endophthalmitis represents a large proportion (approximately 90%) of endophthalmitis cases and refers to instances in which microorganisms directly invade the eye, such as after ocular surgery or a penetrating trauma or as an extension of a corneal ulcer. In terms of postoperative endophthalmitis, exogenous cases are subclassified based on the postoperative time of onset as acute onset (within 6 weeks), chronic onset (more than 6 weeks), or bleb associated (months or years).^1,2,4,6 

Exogenous endophthalmitis, especially postoperative endophthalmitis without systemic disease, may have a poor ophthalmic prognosis but will rarely lead to death. However, comorbidity with one of the systemic conditions described above, especially hematogenous endophthalmitis, can be life-threatening. The incidence of endophthalmitis and its risk factors have been extensively evaluated. However, studies exploring the mortality rates in patients diagnosed with endophthalmitis are rare. In this study, we analyze the epidemiology of and identify potential risk factors associated with mortality in inpatient cases of endophthalmitis. To perform this study, we accessed a nationwide database spanning 14 years. 

Methods

Research Database

The National Health Insurance (NHI) program in Taiwan was launched in March 1995. The NHI covers more than 99% of the population of Taiwan (approximately 23 million people), including all forms of outpatient, inpatient, and emergency health care services. The NHI Research Database (NHIRD) is derived from the claims data of the NHI program and is available to the public in electronic format for research purposes. Data related to basic parameters, such as gender, age, admission days, and diagnosis at discharge, can be obtained in the NHIRD. This database is one of the largest and most comprehensive nationwide population databases in the world. 

Study Participants

This is a cross-sectional study. We used the NHIRD to identify inpatients with a discharge diagnosis of endophthalmitis according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code (360.00, 360.01, and 360.02). We also used ICD-9-CM codes to identify, differentiate, and analyze comorbidities in endophthalmitis patients. We replaced the Charlson comorbidity index (CCI) with the CCI_R (CCI excluding DM, hypertension, renal disease, pneumonia, liver disease, and tumor) because the removed diseases were also variables in this study. From 2000 to 2013, this study identified 7777 patients diagnosed with endophthalmitis out of 11,019,756 total inpatients. Thirteen patients were excluded because they were younger than 18 years old and/or of unknown gender. For patients with multiple admission events, we analyzed variables based on the most recent admission information. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. 

Ethical Considerations

The NHIRD encrypts patient personal information to maintain privacy. Patient consent is not required to access NHIRD data. The Institutional Review Board of the Tri-Service General Hospital approved this study and waived the consent requirement (TSGHIRB No. 2-105-05-082). 

Statistical Analysis

Continuous variables are expressed as the mean ± SD. Normally distributed continuous data related to survival and mortality were compared between the groups using Student's t-test. Categorical variables were compared using either Pearson's χ² tests or Fisher's exact tests. The latter were used in rare events where the expected result was below 5%. Variables were tested using the Firth logistic regression model, and statistical significance was defined as P < 0.05. Several risk factors are presented as odds ratios and 95% confidence intervals from the logistic regression analysis. All statistical analyses were performed using SPSS software, Version 22.0. 

Results

From 2000 to 2013, this study identified 7777 patients who were diagnosed with endophthalmitis in Taiwan out of 11,019,756 hospitalized patients (31,340,284 events). Thirteen patients were excluded because they were under 18 years old or of unknown gender. Of the remaining 7764 cases, 7689 individuals (99.03%) survived, while 75 individuals (0.97%) died (Fig. 1). 

Figure 1

 

A flowchart indicating the process used to select the study sample from the NHIRD in Taiwan.

The trends observed in prevalence and incidence rates in endophthalmitis inpatients between 2000 and 2013 were stable. Compared with female patients, male patients had a slightly higher prevalence (Fig. 2) and a slightly higher incidence rate (Fig. 3). However, there was no difference in inpatient mortality rates between males and females. 

Figure 2

 

Prevalence (per 100,000) of inpatient endophthalmitis from 2000 to 2013 in Taiwan.

Figure 3

 

Incidence (per 100,000) of inpatient endophthalmitis from 2000 to 2013 in Taiwan.

As shown in Table 1, the proportion of male patients was higher than that of female patients (58.09% vs. 41.91%). The mean age was 63.57 ± 15.72 years old in the inpatient cases and 65.23 ± 16.16 years old in the patients who died (the mortality subgroup). This difference was not significant (P = 0.359). The proportion of patients >65 years old was the highest at 55.19% and was followed by the 45- to 64-year-old group (30.74%) and the 18- to 44-year-old group (14.06%). The tendency toward older age in the full group of inpatients with endophthalmitis was also observed in the mortality subgroup in which the largest proportion of patients was in the >65-year-old group (54.67%, 41/75), followed by the 45- to 64-year-old group (34.67%, 26/75), and the 18- to 44-year-old group (10.67%, 8/75). However, the mortality rate was highest in the 45- to 64-year-old subgroup (1.09%) and lowest in the 18- to 44-year-old subgroup (0.73%). Regarding insured premiums (new Taiwan dollars [NT$]) in the full group of endophthalmitis inpatients, most were in the <18,000 group (98.40%), followed by the 18,000 to 34,999 group (1.48%), and >35,000 group (0.12%). The differences in survival and mortality among these groups were statistically significant (P = 0.005) (Table 1). 

Table 1

 

Characteristics of Endophthalmitis Inpatients

 

In the present study, we replaced the CCI with CCI_R (CCI excluding DM, hypertension, renal disease, pneumonia, liver disease, and tumor) since the removed diseases were variables. The CCI_R score was 0.07 ± 0.41 points for the full inpatient group, and there was a significant difference between the survival (0.03 ± 0.41) and mortality (0.24 ± 0.59) subgroups (P < 0.001). 

There was no significant difference between the survival and mortality subgroups in factors such as the season in which the patient was hospitalized, the location in Taiwan and the urbanization level. Out of all of the identified endophthalmitis inpatients, most were treated in a medical center (73.83%), followed by a regional hospital (21.65%) and a local hospital (4.52%). However, a higher proportion of the mortality group was treated in a local hospital (2.63%), followed by a regional hospital (1.76%) and a medical center (0.65%) (P < 0.001). The average length of hospitalization was 16.97 ± 37.88 days, and the duration was longer (30.72 ± 41.95 days) in the mortality group (P = 0.002). The median length of hospitalization was 11 days. The mean medical cost in NT$ was 86,462.75 ± 143,243.66, and it was higher in the mortality group (274,740.01 ± 289,185.84 NT$) (P < 0.001). The median medical cost was 49,200 NT$ (Table 1). 

The epidemiological characteristics of the patients were compared (as with or without) based on the presence of systemic comorbidities, and the results revealed that the adjusted odds ratio (AOR) was significantly higher when the comorbidity was renal disease (AOR 2.864, P = 0.001), septicemia (AOR 8.886, P < 0.001), pneumonia (AOR 2.072, P = 0.030), or a tumor (AOR 7.437, P < 0.001). Comorbidity with DM was associated with a 0.5-fold lower AOR (P = 0.026). However, comorbidity with hypertension, depression, anxiety, hyperlipidemia, thyrotoxicosis, liver disease, and injury was not significantly different according to the odds ratios (all P > 0.05) (Table 2). 

Table 2

 

Characteristics of Endophthalmitis Inpatient Prognoses With the Firth Logistic Regression Model

 

Discussion

Although endophthalmitis may not be a major or direct cause of patient deaths, mortality can result from poorly controlled underlying comorbidities and systemic infection. Vaziri et al.⁵ conducted a 5-year retrospective study that revealed the odds of a diagnosis of endogenous endophthalmitis among patients with hematogenous infections was higher in patients who also had endocarditis, bacterial meningitis, lymphoma/leukemia, internal organ abscess, DM with ophthalmic manifestations, or longer hospital stays. However, few studies have explored the associations between risk factors and mortality in inpatients with endophthalmitis.^5,7 To the best of our knowledge, the present study is the first such large population-based study to use a nationwide database to evaluate the epidemiology and predictors of inpatient mortality among endophthalmitis patients in a Chinese population. The data included in this analysis covered 14 years. 

In the present study, the mortality rate was 0.97% (75 deaths/7764 endophthalmitis cases). A 4% mortality rate was reported by Jackson et al.⁶ in a systemic review of 342 cases. However, they evaluated only endogenous cases, and it is likely that their endogenous group had a higher mortality rate because affected patients had more comorbidities and more critical medical conditions than were observed in the exogenous group. In terms of age, the mean age of the endophthalmitis inpatient group evaluated in the present study was 63.57 years old, which is older than the mean age of the patients in the endogenous endophthalmitis review study (52 years old) by Jackson et al.⁶ Nonetheless, in a systemic review of the China National Knowledge Infrastructure database that included 57 articles (403 eyes), the mean age of Chinese patients with endophthalmitis following the development of a cataract was 68.8 years old.⁸ Based on these studies, we assumed that the mean age would be younger in the endogenous group than in the exogenous group. We combined exogenous and endogenous entities, and the results shown include both groups. 

In this study, there were more males than females in the full group of endophthalmitis inpatients, which was similar to other reviews.^6,9–12 Nevertheless, this gender difference was not statistically significant. A higher mortality rate was noted in the high insured premiums group than in the low insured premiums group (12.5% in >35,000 group vs. 0.98% in <18,000 group, P = 0.005). This elevated mortality rate may be explained by the higher comorbidity found in the high insured premiums group (mean CCI scores ± SD was 1.11 ± 1.537 in the >35,000 group vs. 0.54 ± 1.335 in the <18,000 group) that resulted in a more severe disease status. 

Comorbidity with hypertension, depression, anxiety, hyperlipidemia, thyrotoxicosis, liver disease, or injury did not influence the incidence of mortality (Table 2). However, inpatients with renal disease had a 2.86-fold higher risk of mortality, which may be because patients with renal disease, especially those with end-stage renal disease being treated with hemodialysis, are at a higher risk of experiencing a systemic infectious episode or exhibiting poor immunologic conditions. The risk of mortality was higher in inpatients (8.89-fold) with septicemia than in those with renal disease. In these patients, endophthalmitis may not be the only cause (or even the major cause) of septicemia. In fact, endophthalmitis is often an end-organ complication rather than the source of septicemia. Septicemia usually indicates decompensated systemic immunity and is associated with weakened host defenses, which makes it easier for pathogens to spread through the bloodstream. Therefore, the mortality rate is higher in patients with this critical condition.^2,4,13 

In the study by Vaziri et al.,⁵ the odds of a diagnosis of endogenous endophthalmitis were not significantly affected by the presence of bacterial pneumonia or intestinal infections, perhaps indicating a low likelihood of hematogenous spread.⁵ Nevertheless, the results of the present study showed that inpatients with pneumonia had a 2.07-fold higher risk of mortality. This result could be interpreted as indicating that even though pneumonia does not increase the incidence of endophthalmitis, severe pneumonia with respiratory failure does increase its mortality rate. Additionally, the AOR was 7.44-fold higher in patients with a tumor than in those without a tumor. This result is consistent with the notion that malignancy itself may compromise host defenses against the hematogenous spread of pathogens. Furthermore, most of these patients were treated with systemic chemotherapy, which caused a profound and catastrophic immunocompromised status, and these factors were associated with a higher mortality rate. In this study, we also calculated the CCI_R and analyzed the relationship between CCI_R scores and ORs in inpatient mortality. We found that CCI_R scores were higher in the mortality group. It is reasonable to presume that CCI_R scores are generally higher in older patients and in those with conditions with more advanced severity, which may explain the relationship between high CCI_R scores and a higher mortality rate. 

One interesting result of the present study was that DM was associated with a lower risk (0.5-fold) of inpatient mortality. DM with poor controlled glycemic status is associated with poor wound healing because the condition interferes with leukocyte chemotaxis and impairs phagocytosis functions.¹⁴ DM also consequently increases inflammation and the rate of infection, therefore increasing the incidence of endophthalmitis. Sheu et al.¹⁵ conducted a 20-year retrospective review of 42 patients (53 eyes) in southern Taiwan and found that DM was a significant risk factor for the development of endogenous endophthalmitis and poor visual outcomes in patients with a Klebsiella pneumoniae (K. pneumoniae)-infected liver abscess.¹⁵ Nevertheless, few studies have explored the relationship between DM and the risk of mortality in endophthalmitis, especially in Asian populations. In this study, endophthalmitis inpatients comorbid with DM decreased the OR in mortality. One possible explanation for this effect is that long-term DM, especially in patients with poor glycemic control, is frequently complicated by vascular or infectious diseases, such as nephropathy, pneumonia, and septicemia. When the effects of these comorbidities on mortality were compared in total DM patients, DM was found to act as a protective-like prognostic factor. Another reason may be that in Asian populations of endophthalmitis patients with DM, the most frequently identified pathogen is K. pneumoniae, which hematogenously spreads from a liver abscess. In affected patients, a complete picture of the disease was obtained, and treatments, including appropriate antibiotics, and/or surgical approaches, such as intravitreal injection, vitrectomy, and the drainage of liver abscess, were performed. However, in patients without DM, endophthalmitis may be caused by different pathogens, including more invasive microorganisms, such as methicillin-resistant Staphylococcus aureus (MRSA), bacilli, and fungi, depending on the primary source of the infection.^2,4,16 In a review article, Han¹⁷ discussed the association of hepatic abscess from K. pneumoniae with DM and septic endophthalmitis; the author concluded that septic endophthalmitis in patients with a liver abscess due to gram-negative organisms other than K. pneumoniae is associated with a poor outcome.¹⁷ Therefore, management of these patients can be more difficult, resulting in a higher rate of mortality. Briefly, when we treat a systemic infectious patient, especially one without a history of DM, it is important to pay attention to the initial endophthalmitis symptoms and signs, including redness, lid swelling, pain, rapid deterioration in vision, ciliary injection, and uveitis. If we can obtain an early diagnosis in these patients and promptly and intensely treat them, we may improve their prognosis. 

In the present study, we also analyzed the AORs of mortality in endophthalmitis inpatients according to factors including the season at the time of admission, outpatient locations, and urbanization levels. There was no significant difference between patients who died or did not die according to these factors. In terms of the level of care, most of the included endophthalmitis cases were treated in medical centers (73.83%), while the remainder were treated in regional (21.65%) and local (4.52%) hospitals. However, the proportion of patients in the mortality group was reversed, with the highest proportion found in the local hospital group (2.63%, P < 0.001). The crude OR revealed a 0.252-fold lower risk of mortality in the medical center group than in the local hospital group (P < 0.001). Although these results may not be reliable, these data could indicate that mortality is lower in medical centers, which generally offer a more complete spectrum of services. Therefore, critical and severe patients who are treated in these facilities are more likely to receive better healthcare interventions and care, resulting in a lower mortality rate. 

The duration of hospital stays and medical costs were significantly higher in the mortality group. This finding is consistent with the widely accepted notion that we spend more effort and have higher expenditures on critical cases, some of whom eventually die. 

The present study has some limitations. First, this study identified a total of 7764 inpatients who were diagnosed with endophthalmitis and entered into the NHIRD over a 14-year period. However, we did not differentiate endogenous from exogenous patients, and these groups can present different characteristics. For example, endogenous cases present with more comorbidities and more severe systemic signs, whereas exogenous cases are more often the result of an accident, such as trauma or a postoperative complication. Second, the NHIRD does not offer detailed clinical information, such as disease severity, clinical manifestations, and laboratory findings. Therefore, we could not perform further analysis or determine the cause of death without using a chart review method. Third, visual impairment is a major complication of endophthalmitis. However, we could not obtain data for this condition from the NHIRD since there are no precise ICD-9 codes corresponding to visual impairment. Fourth, mortality cases may influence the inpatient length of stay and medical cost results, resulting in either underestimation or overestimation. Thus, we should keep in mind that the length of stay and medical costs were death-sensitive variables when we interpreted the results in this study. Finally, in the present study, we evaluated only endophthalmitis occurring in inpatients and did not include patients who died after hospital discharge. However, there are fewer such out-of-hospital deaths than inpatient deaths in endophthalmitis. 

Conclusions

In conclusion, we searched a nationwide research database and evaluated 7764 inpatients cases of endophthalmitis that occurred in Taiwan from 2000 to 2013. We found that the mortality rate in endophthalmitis inpatients was 0.97%, the mean age was 63.57 years old, and there was no difference in gender proportions between the groups. Furthermore, the present study identified several predictors of mortality in endophthalmitis occurring in inpatients. These predictors included renal disease, septicemia, pneumonia, neoplasia, a greater burden of comorbidity (especially catastrophic illness), longer duration hospital stays (more than 11 days), and higher medical costs. However, DM resulted in a decreased OR in inpatient mortality. In systemic infectious patients without a history of DM, care must be taken regarding the initial endophthalmitis signs and symptoms to prevent the further progression of the disease. Finally, we propose that nationally representative data such as that obtained from this database provide useful information regarding predictors of mortality and risk factors in endophthalmitis inpatients, especially in Asian populations. 

Acknowledgments

This study is based on data obtained from the NHIRD, which is provided by the Bureau of National Health Insurance, the Department of Health, Taiwan and managed by the National Health Research Institutes. 

Supported by the Tri-Service General Hospital, Taipei, Taiwan, Republic of China. 

Disclosure: T.-H. Weng, None; H.-C. Chang, None; C.-H. Chung, None; F.-H. Lin, None; M.-C. Tai, None; C.-H. Tsao, None; K.-H. Chien, None; W.-C. Chien, None