Aim To investigate the risk factors and comorbidities associated with ethambutol-induced optic neuropathy (EON).
Method Using the Taiwan Longitudinal Health Insurance Database, we conducted a study within a nationwide representative cohort of patients treated with EMB. We identified 231 patients newly diagnosed with EON between 2000 and 2008, and 924 control subjects. Adjusted OR by estimating the risk of EON in relation to comorbidities and EMB prescription protocol was determined.
Results Compared with the control group, EON patients were at risk with older age, hypertension (adjusted OR=1.62, 95% CI 1.16 to 2.26) and renal diseases (without end-stage renal diseases (ESRD), adjusted OR=2.11, 95% CI 1.02 to 4.35; with ESRD, adjusted OR=3.73, 95% CI 1.79 to 7.74). Patients with an EMB prescription duration longer than 3 months were not at elevated risk compared with those whose prescription less than 3 months (OR=1.35, 95% CI 0.99 to 1.83, adjusted for age, sex, hypertension and renal diseases). Patients whose average daily dose was greater than 1200 mg, compared with the other two groups (800∼1199 mg, less than 800 mg) were not at increased risk for EON.
Conclusions Age, hypertension and renal diseases are risk factors for EON in the Taiwanese population.
- Eye (Globe)
- Optic Nerve
Statistics from Altmetric.com
Although ethambutol (EMB) is well known to be the most common cause of toxic optic neuropathy,1 ,2 it is still commonly used in medical practice due to increasing drug resistance in tuberculosis treatment.1 ,3 The incidence of ethambutol-induced optic neuropathy (EON), its clinical characteristics and risks factors and its dosage information, have been poorly described and characterised in many previous studies.2–6 However, the causes of EON and its risk factors remain to be determined. Some previous studies showed that EON is dose-related and reversible,7 but others revealed a controversial result.8–12 Because of these ambiguities, the reported incidence of the toxicity varies widely across studies, ranging from 0.5% to more than 35%.4 ,13–15 Unfortunately, with the increase in the number of cases reporting irreversible EON,,4–6 an urgency exists to identify the potential risk factors associated with EMB. To better understand the risk factors related to EON in our population, we conducted this study using an 8-year population-based dataset from the Taiwan National Health Insurance (NHI) Programme.
The NHI programme offers healthcare coverage to more than 99% of the 23 million people. Electronically submitted reimbursement claims data have been available since 1996, and are sent to the National Health Research Institute which collaborates with the NHI to construct the NHI Research Database.16 The Longitudinal Health Insurance Database was established for research purposes. This dataset, comprising one million beneficiaries randomly selected from all people enrolled in the NHI between 1996 and 2000, includes files of registration, ambulatory care claims, inpatient claims and prescription drugs, and provides information on healthcare utilisation.
Cases and controls
This study consists of 11 753 subjects who were dispensed EMB between 1996 and 2008. The cohort was comprised of patients who were 20 years of age or older and who were diagnosed with EON between 2000 and 2008 on either an inpatient or an outpatient basis. All cohort participants were identified using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) code 377. A control group was randomly selected from the remaining subjects who did not experience EON but were on EMB during the specified time period. This resulted in a case-to-control ratio of 1 : 4. The index date of the case subject was the date of their EON diagnosis. For each control subject, we assigned an index date frequency matching the index year and the month associated with each of the case subjects. Study subjects with missing information on sex and birth date were excluded. Finally, a total of 231 patients with EON (cumulative incidence, 2% (231/6369); incidence density: 36.3 per 1000 person-years) and 924 control group subjects were included.
We scanned diagnosis information on outpatient and inpatient claims before the index date to identify some comorbidities, including diabetes mellitus (ICD-9 250.XX), obesity (ICD-9 278.0, 278.00, and 278.01), hypertriglyceridaemia (ICD-9 272.1), alcoholism (ICD-9 303, 305.00, 305.01, 305.02, 305.03, and v11.3), gallstones (ICD-9 574.00, 574.01, 547.10, 574.11, 574.20, and 574.21), hepatitis B (ICD-9 V02.61, 070.20, 070.22, 070.30, and 070.32), hepatitis C (ICD-9 V02.62, 070.41, 070.44, 070.51, and 070.54), hypertension (ICD-9 401–405), renal diseases (ICD-9 585, 586, 588.8, and 588.9), systemic lupus erythematosus (ICD-9 710.xx), Reiter's syndrome (ICD-9 711.1x), rheumatoid arthritis and other inflammatory polyarthopathies (ICD-9 714.xx), palindromic rheumatism (ICD-9 719.3x), polymyalgia rheumatica (ICD-9 725), and malignant illness (ICD-9 140–208). To assess whether the risk of EON increases in patients with advanced-stage renal diseases, we further classified this group into two subgroups (end-stage renal diseases, ESRD and non-ESRD). Patients with ESRD were identified using the Catastrophic Illness Registry. In the NHI programme, patients with ESRD were eligible to apply for a catastrophic illness certification, which allows for the waiver of copayments for medical care related to ESRD. In both the EON case group and the controls, prescription duration and average daily dosage of EMB were calculated from the date of the participant's first prescription until the index date. Average daily dose was calculated by dividing total number of milligrams prescribed by total number of days supplied during the year before the index date. According to the standard prescription guideline for EMB,17 we stratified the average daily dosage into three categories: (1) less than 800 mg, (2) 800–1199 mg, (3) greater than 1200 mg. The compliance in these patients is assessed under directly observed therapy by healthcare personnel.
We compared the differences in sex, age and comorbidities between case group and control group using Pearson's χ2 test. The logistic regression model, which estimated the OR and a 95% CI, was used to evaluate the association between age, sex, comorbid disease and the risk of EON. The logistic regression analyses did not include the comorbidities in which the prevalence rates were similar between case group and the controls, or the comorbidities in which the prevalence rates were very low in both groups. We did not select the comorbidities for inclusion relying on p values due to the small number of cases with comorbidities. We further examined whether patients with coexisting hypertension, diabetes and renal diseases were at a greater risk of EON. ORs for optic neuropathy in relation to duration of EMB prescription and average daily dosage were calculated as well. All analyses were performed using SAS statistical software (V.9.1 for Windows; SAS Institute, Inc, Cary, North Carolina, USA). We used a two-tailed test for all analyses and the significance level was set at a p value <0.05.
In both the EON group (case) and the non-EON group (control), patients taking EMB were more likely to be male (table 1). The patients in the EON group (65.4±15.9 years) were older than the controls (60.3±19.6 years). Compared with the control group, the EON group was more likely to have hypertension (54.6% vs 38.6%) and renal diseases (non-ESRD, 5.6% vs 2.5%; ESRD, 6.9% vs 1.7%, p<0.0001).
In the logistic regression models adjusted for age, sex, diabetes, hypertension, renal diseases and malignant illness, the OR for EON in patients aged 40–64 years, compared with patients in the 20–39-year age range, was 1.76 (table 2, Model 2). The adjusted OR increased to 1.93 for patients aged 65 years and older. Hypertension was associated with an increased risk of EON (OR=1.62). Additionally, patients with renal diseases also showed an elevated risk for EON compared with subjects without renal diseases. For patients who had renal diseases but without ESRD, the adjusted OR was 2.11; the adjusted OR rose to 3.73 for patients with ESRD. However, diabetes was associated with a decreased risk of EON. The risk was not statistically significant in the age-adjusted and sex-adjusted model (Model 1, OR=0.71), but the risk became significant after an adjustment was made for other comorbidities (Model 2, OR=0.58).
Table 3 indicates the adjusted ORs of EON in patients with coexisting diabetes, hypertension and renal diseases compared with those without these diseases. The OR for patients with hypertension alone was 1.79 and increased to 3.84 for hypertensive patients with renal diseases. The OR for patients having all three of these comorbidities was 3.94. For renal diseases alone, the OR was 2.42, but it was not statistically significant.
Table 4 presents the OR of EON in relation to duration of the EMB prescription. Patients whose EMB prescription duration was longer than 3 months were found to be at a mild elevated risk of EON compared with those less than 3 months (OR=1.38, 95% CI 1.02 to 1.86, Model 1, adjusted for age and sex). However, when additionally adjusted for hypertension and renal diseases, there is no increased risk (OR=1.35, 95% CI 0.99 to 1.83).
Table 5 shows the result of OR for optic neuropathy in relation to average daily dosage of EMB prescription. For all cases treated with EMB, relative to the group whose average daily dose was less than 800 mg, the other two groups were not with increased risk for EON. The same result was noted in the cases with renal diseases. Mild increased risk for EON was only noted in the group whose average daily dose was between 800 and 1199 mg in the cases without renal diseases (OR=1.46, 95% CI 1.04 to 2.06).
This study used population-based data to understand the comorbidities associated with EON in Taiwan, which is one of the few studies investigating the roles of multiple systemic factors in EON in a large Chinese cohort. Our results show that patients with EON were found to be older and to have hypertension and renal diseases compared with the controls. Our result is in agreement with the general belief that advanced age and renal diseases are the two most common risk factors for EON.2–6 ,13 ,14 In one recent meta-analysis of 70 EON cases, the authors noted that the majority of patients were over 40 years of age, and 40% of them were over 65 years,3 which is consistent with our result (57.6%, over 65 years old). However, when the authors further considered the CDC's statistics on the age distribution of tuberculosis cases, they found that the number of cases of tuberculosis decreased with increasing age after peaking in the 25–44 age group.3 They postulated that age was an independent risk factor for EON based on the thought that renal function decreases related to aging.3,18 Although renal tubular function is known to decrease with age,18 the aging effect of EMB clearance is not yet known. Compared with that study, our current logistic analysis model could explain why age should be a strong risk factor for EON.
Furthermore, our analysis identified some other interesting findings. First, hypertension was associated with an increased risk of EON (OR=1.62), which has seldom been reported.2–6 ,13 ,14 Second, diabetes was associated with non-statistically significant reduction in risk of EON in the age- and sex-adjusted model, and the association became statistically significant after further adjusting hypertension and renal disease. The role that diabetes plays in EON needs to be further elucidated in a future study. Third, patients with renal diseases were also at an elevated risk for EON compared with subjects without renal diseases. The adjusted OR for patients with renal diseases, but without ESRD, was 2.11, and this rose to 3.73 for patients with ESRD. This important result confirms the previous belief that poor renal function is an important risk factor for EON.2–6 ,13 ,14
Because hypertension and renal disease are both risk factors of EON, we further examine the joint effects of these comorbidities on EON risk. To assess the influence of diabetes on the associations, the analysis was also stratified by diabetes. The adjusted OR for patients with hypertension only was 1.79, and for patients with renal diseases only it was 2.42 (not statistically significant). The adjusted OR for patients with hypertension only was 1.79 and increased to 3.84 for those with coexisting renal disease. The OR was at similar strength for patients who were also comorbid for diabetes. This study is the first to notice that hypertension is a potential aggravating factors for EON in patients with renal diseases.3 ,5
Conflicting results about the treatment protocol of EMB on EON were noted.12 ,13 One recent meta-analysis study has reported that the duration and dose of EMB were positively correlated with risk of EON.13 In another study in Korea,12 they reported that renal dysfunction and daily dose of EMB, but not duration of EMB treatment, seem to be related to development of EON. To clarify this important issue in our population, we further analysed the EMB dosing regimen, including the prescription duration and average daily dosage. Interestingly, our result reveals that both prescription duration and average daily dosage do not have significant association with the risk of EON. We believe this most likely represents an idiosyncratic reaction of EMB in our Chinese population. Furthermore, we found the majority of EMB-treated cases (about 88%) were, on average, less than 1200 mg/per day. According to the treatment guideline of tuberculosis,17 recommended dose was standardised for three or four body weight bands. It is a great pity that the body weight data were not available in our dataset; therefore, we failed to understand the real daily dose based on weight. This is a major limitation in this study compared with others, and constrains a direct comparison with previous studies because of different study design and statistical method. However, this result could be a good basis for future prospective studies in EON.
Although the results are quite meaningful, some limitation should be mentioned. First, the EON diagnoses were totally reliant upon claims data (ICD-9 coding from clinicians), which may be less accurate than diagnoses carried out individually through a standardised procedure. Second, other selection biases also inherit in this kind of retrospective study. Because the vision and body weight data were not available in our health insurance database, we were unable to understand the real dose effect on the severity of vision impairment in each patient based on body weight. Even so, this dataset has good sample randomisation.19 Third, to avoid misdiagnosis for the case group, we ensured that all study cohorts should have EON diagnoses in consensus with concurrent and continuous intake of EMB. Fourth, because we could not determine compliance in this retrospective study, however, comments on how compliance is assessed in these patients (directly observed therapy, etc) would be useful to have. Additionally, we excluded any tuberculosis case coded with any optic neuropathy before EMB use.
In summary, age factor, hypertension and renal diseases were found to be important risk factors for EON. The important issue of dosing regimen of EMB on EON should be clarified in future studies. We hope that our current work would help patients avoid this vision-threatening disease.
Contributors Dr CH-Y: conception and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, critical revision of the manuscript for important intellectual content, obtaining funding. Dr S-WL: conception and design, analysis and interpretation of data, critical revision of the manuscript for important intellectual content. C-HM: statistical expertise, conception and design, critical revision of the manuscript for important intellectual content. Dr P-CC: statistical expertise, conception and design, critical revision of the manuscript for important intellectual content, obtaining funding. Dr I-JW: critical revision of the manuscript for important intellectual content, supervision.
Financial disclosure(s) The authors have no commercial or proprietary interest in any of the materials discussed in this article. This work was supported by Taiwan National Sciences Council (NSC 98-2621-M-039-001), Department of Health Clinical Trial and Research Center of Excellence (DOH100-TD-B-111-004), and the China Medical University Hospital (1MS1).
Competing interests None.
Ethics approval This kind of database study, provided by our health insurance research unit, does not need the IRB evaluation in our school.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.