Contact Dermatitis after Prescription of an Ophthalmic Ointment Containing Fradiomycin Sulfate: A Retrospective Database Study Using Japanese Health Insurance Claims Data

Fradiomycin sulfate (neomycin)/methylprednisolone (F/M) (Neo-Medrol EE ointment, Pfizer Japan Inc., Tokyo, Japan) has been prescribed for the treatment of ophthalmic inflammation caused by infection, external ear eczema and dermatitis, and post-operative treatment in the otorhinolaryngology field in Japan for more than 50 years. Fradiomycin sulfate is an aminoglycoside antibiotic that is known to cause allergic contact dermatitis (CD) in some patients [1, 2]. In addition to F/M, fradiomycin sulfate/betamethasone sodium phosphate (F/B) (Rinderon-A ointment, Shionogi & Co., Ltd., Osaka, Japan) ophthalmic ointments with indications nearly identical to those of F/M are also available.

Over the past 3 decades, CD cases showing a potential causal relationship with F/M or fradiomycin have been reported [3‐21]. However, reliance on case reports has resulted in a lack of studies on the incidence of CD or systematic investigations of the causal relationship with a specific drug in a large population in Japan.

The aim of this study, therefore, was to investigate the incidence of CD in patients with prescriptions for F/M as compared with that for F/B and two other drug groups of ophthalmic ointments (those containing either an antibiotic or a steroid), using the MinaCare Co. Ltd. healthcare database (MinaCare HDB), a commercial insurance claims database. The MinaCare HDB includes the data of approximately 2 million individuals with medical and pharmaceutical claims data received from employment-based health insurance companies and covers a wide range of age groups.

In this study, an insurance claims database was used to assess the incidence rate of CD requiring medication after the filling of prescriptions for F/M. To achieve this aim, we conducted a comparative assessment of four targeted drug groups in actual clinical settings. The four drug groups were ophthalmic ointment formulations of antibiotics, steroids or a combination of the two. No new product or new indications for these drug formulations have been launched or approved in Japan since 2013, and the study was conducted using the database containing the claims data for prescriptions filled before 2013. The medical conditions for which prescriptions for the investigational drugs had been filled involved predominantly the treatment of ophthalmic conditions in all four groups. The age and sex distributions of the study subjects were similar in the four groups (Fig. 1). Regarding the medical conditions, major diagnoses at the initial prescription during the investigational period were similar in the F/M and F/B groups. A slight difference was seen between the antibiotic group and the other three groups, which might have affected the incidence of CD. Under the conditions of such a difference being taken into consideration in the comparison between the antibiotic group and the other groups, we think that further analyses using the data set would be valid.

Based on this database study, there were no clear differences in the incidence rate of CD during the 12-month follow-up period between the F/M and F/B groups. A sensitivity analysis showed the same tendency in all of the follow-up periods (Fig. 3). There is a report showing that F/M ranks high as a topical steroid (including combination) drug with positive patch test results (seven out of 55 cases) [25]. In the present retrospective study, based on the claims database, the incidence rates of CD were similar in the F/B and steroid groups.

The incidence rate of CD was higher in the F/M, F/B, and steroid groups than in the antibiotic group, by 0.03–0.04. We speculated that this might have arisen from the difference in the characteristics of the patients, especially their skin conditions, which influence the selection of each drug. For example, CD generally develops more frequently at eczematous sites, and the drug groups including steroids are regarded as treatment options for eczema and are thus used more frequently than antibiotics to treat eczematous lesions. Additional analysis showed that eczema was diagnosed more frequently in the F/M, F/B, and steroid groups (11.4–16.2%) than in the antibiotic group (3.8%) at the initial prescription during the investigational period. Furthermore, the percentage of CD cases with a history of eczema before the diagnosis of CD in the antibiotic group was similar to that in the other groups (F/M 20.0%, F/B 21.2%, antibiotics 15.7%, and steroids 13.5%). Based on these results, we think the lower CD incidence rate in the antibiotic group might be explained by other drugs having been prescribed for eczematous lesions, including CD. Considering the potential bias in drug selection, such as the characteristics of the patients, we advocate not taking this low CD incidence rate into consideration when interpreting the slight difference in rates between the antibiotic group and the other three groups. Furthermore, it merits emphasis that treatment warnings are equally necessary for all investigational drugs reviewed in this study.

As for the demographic information of patients who had filled prescriptions for the investigational drugs, our results showed their age distributions to be similar to that of the entire MinaCare HDB (Fig. 1). However, the results obtained in this study are not applicable to elderly patients based on the subject composition of this database.

The present results showed that the incidence rates of CD in males and females in the antibiotic group did not differ among age groups. On the other hand, the incidence rates of CD in females ≥ 20 years old were higher than the rates in the male groups and in females < 20 years old for the other three investigational drug regimens (Fig. 4). The proportions of females ≥ 20 years old who filled prescriptions for drugs in the F/M, F/B, and steroid groups were higher (68.8–75.0%) than the proportion in the antibiotic group (56.9%). This may suggest a tendency for females ≥ 20 years old to pay closer attention to their skin conditions, and the higher incidence rates of CD in these three drug groups may reflect the higher proportion in the study population of females ≥ 20 years old.

Case reports of CD related to F/M are summarized in Table 3 [3‐21]. Based on the information obtained, the reported number of female patients was larger than that of male patients, and the majority of patients ≥ 20 years old were female, which is similar to the results obtained in this study. Although it is presently unclear why a greater incidence of CD is reported in females ≥ 20 years old who filled prescriptions for drug formulations that contained steroids rather than antibiotics, it is assumed that this group’s interest in cosmetics attributes to their awareness of dermatological care and they are therefore more sensitive to the dermatological symptoms associated with the use of steroids. Furthermore, this could also be explained by an already existing risk factor like an eczematous lesion or CD associated with cosmetics, such as eye make-up products.

The present results showed that the patch tests that should be performed at diagnosis, as clearly stated in the CD treatment guidelines [26], are not performed as recommended (Table 2). This raises the possibility that the patch test was only performed when identification of the causal drug was absolutely necessary.

It was also found that the same prescription was filled repeatedly after the diagnosis of CD had been made, despite the prescribed drug being suspected to have caused the CD (Table 2).

Herein, we have presented the incidence rates of CD caused by specific drugs, based on our analysis of the MinaCare HDB. The MinaCare database reflects the health status of those enrolled in employment-based health insurance and provides a large sample size with a broad age distribution. This database may serve as a reliable tool for post-marketing evaluations of drugs.

The significance of this study is that it explored the incidence of CD associated with treatment in actual clinical practice, based on analysis of prescription records in Japan, as compared to the conventional analysis of case reports for which the antigen had already been identified by patch tests [1, 2, 25]. However, this database study has several limitations. The first limitation is that the incidence rate of CD was calculated only for patients who visited hospitals and were diagnosed by doctors as having CD. There may be a large pool of patients who actually had CD but did not visit the hospital and therefore went undiagnosed and unrecorded. The second limitation is that the information in the claims data in the MinaCare HDB is limited. The severity of the disease, causal relationship of the investigational drug formulations with the disease, and patient adherence to dispensed drugs or the frequency of dispensed drug use by the patients (and possibly by other family members) are not fully documented in the HDB. Regarding the objective of prescribing the investigated drugs, the possibility that the prescriptions for CD include uses for treating other disorders cannot be denied. In addition, the lack of patch testing makes it difficult to prove the direct association with CD. The causal relationships between the investigational drug formulations and CD require primarily evaluation based on whether the disease occurred before or after prescription of the investigational drug regimens. Because the site of CD is not reported in the HDB, there is a possibility of CD unrelated to the investigated drugs observed at sites other than the eyes having been included in the incidence. Since only data associated with the targeted investigational drugs were analyzed, no other factors such as concomitant medications or non-drug allergens that could have caused CD have been taken into account in the present study. According to a report on patch tests using the Japanese Standard series in CD patients, the positive rate at 72 h was very high for metals such as nickel sulfate or cobalt chloride (18–19%), and 2–8% of those examined showed a positive reaction to fragrance mix, paraben mix, p-phenylenediamine-free base or lanolin alcohol [27]. Wearing make-up or accessories, in other words exposure to these allergens, might have led to the higher incidence rate of CD observed in females in this study. In addition, the diagnosis based on the disease codes may not reflect the actual medical records. To overcome such limitations and reduce bias, we took two approaches. We set the washout and follow-up periods, and evaluated the results through a comparison with proper comparator drugs under the same conditions. By taking these two approaches, we believe that the relationships between the dispensed investigational drugs and the diagnoses of CD were appropriately estimated.

In countries such as Germany that have information networks on contact allergies [28], the prevalence and risk assessments of contact allergy have been made on the basis of the surveillance results for topical drugs [29‐32]. In Japan, however, no such system currently exists and it is difficult to obtain nationwide surveillance information. We therefore believe that the results of this study provide valuable large-scale real world data, despite the limitations mentioned above.

We assessed the incidence of CD requiring medication after prescriptions for topical drugs had been filled in actual clinical settings, using the MinaCare HDB. We found that the incidence of CD that required any form of treatment after the filling of a prescription for F/M was similar to that for F/B and ophthalmic ointments containing a steroid, and slightly higher than that for an antibiotic. The study also revealed that the investigational drugs were repeatedly dispensed after the diagnosis of CD. It is critical for primary care physicians to have a better understanding of the risk of CD due to these prescribed ophthalmic ointments, and they should take appropriate measures such as stopping the use of the possible prescribed causative drug or antigen when encountering patients who have developed dermatitis. Additionally, as recommended by the guidelines, identification of the causative drugs and ingredients using patch tests after CD has developed is important for avoiding further exposure to the allergens [26, 33].