Health service use and costs associated with fluoroquinolone‐related tendon injuries

Abstract The aim of this study was to assess costs and health service use associated with tendon injuries after the use of fluoroquinolone antimicrobials in Finland during 2002–2012. This retrospective observational study included data from the Finnish Pharmaceutical Insurance Pool's pharmaceutical injury claims. In total, 145 compensated claimants aged ≥18 years presenting tendon injuries after the use of fluoroquinolones (FQs) were included in the study. Outcomes of interest were the number of outpatient visits to primary, secondary, tertiary, and private healthcare services, hospital days, rehabilitation and their costs. Regression models were used to analyze the impact of patient characteristics on hospital days, as well as the relationship between patient characteristics and tendon ruptures. Direct costs of a tendon injury averaged 14,800€ and indirect costs were estimated to be 9,077€ for employed claimants. Fifty‐one percent of the claimants were hospitalized, with an average duration of 21 days. Hospitalization was the costliest form of health service use with an average of 9,915€ per hospital episode. Hospital days and direct costs increased with the severity of the injury. Tendon ruptures, in particular bilateral ruptures, required substantially more hospital days and their direct costs were significantly higher than those of uncomplicated tendinitis. Concurrent use of oral corticosteroids and increasing age were associated with a higher likelihood of tendon ruptures. Although rare, FQ‐related tendon injuries can result in considerable costs and health service use. Medical staff should remain vigilant when prescribing FQs, especially in groups at increased risk for tendon injuries.

and neuropathies. Consequently, the European Medicines Agency (EMA) and the U.S. Food and Drugs Administration (FDA) have recommended several restrictions on their use. 3,4,5 Tendon ruptures and tendinitis are rare but debilitating and possibly costly AEs that have been strongly associated with the use of FQs. According to a recently published cohort and nested case-control study, the excess risk of a tendon rupture due to fluoroquinolones is about 3.7 cases per 10,000 person-years. 6 Signs of tendinitis include tendon swelling, irritation, and moderate-to-severe pain. Tendon ruptures, on the other hand, usually cause severe pain, weakness, and deformity of the affected tendon. Tendon injuries are treated with immobilization and/or corrective surgery. 7 Both tendinitis and tendon ruptures can either heal completely or cause permanent damage to the tendon.
The treatment of these serious AEs often requires both sick leave and health service use, which, in turn, cause potentially significant costs for patients and society. Previous research has shown that assessments of FQ-related AE costs are scarce even though hospitalization is frequently required to treat AEs. 8 The specific molecular pathway causing FQ-related tendon injuries remains unidentified. The biological antimicrobial mechanism of FQs is based on targeting bacterial DNA gyrase (DNA topoisomerase), 9 and accordingly, a possible mechanism of AE is that FQs not only target bacterial DNA but also human DNA. A recent study has proposed that most FQ-related AEs could be explained by mitochondrial DNA effects due to the loss or inhibition of type II topoisomerase Top2β. 10 In addition, there is evidence of a connection between FQs and other collagen-associated AEs, such as aortic aneurysms, aortic dissections, and retinal detachments, in addition to tendon injuries. 11 Previous research has shown that concurrent use of corticosteroids, an age of over 60 years, and chronic kidney disease are the risk factors for developing FQ-related tendon injuries. 6,12 The aim of this study was to assess costs and health service use associated with tendon injuries relating to FQ antimicrobial use in Finland during 2002-2012. Although FQ-related tendon injuries were first reported in 1983, 13 as far as we know, the costs associated with them have previously not been studied.

| MATERIAL S AND ME THODS
This retrospective observational study analyzed data obtained from the Finnish Pharmaceutical Insurance Pool's pharmaceutical injury claims. In Finland, patients, who suffer a pharmaceutical injury are entitled to compensation, if a causality can be detected between the pharmaceutical and the injury. A pharmaceutical injury is a legal term defined as any bodily illness or injury or a psychiatric disease likely to result from a pharmaceutical taken by the injured party. 14 Pharmaceutical injury compensations are determined by applying the provisions contained in Finland's Tort Liability Act and the guidelines issued by the Traffic Accident Board.
The costs of FQ-related tendon injuries were evaluated from the Insurance Pool's compensated insurance claims from 2002 to 2012.
FQ-related tendon injury costs were borne by the Social Insurance Institution of Finland, municipalities, employers, patients, and the Finnish Pharmaceutical Insurance Pool. Of these payers, the latter is a secondary insurance system and only compensates patient costs relating to pharmaceutical injuries, which are not covered by other institutions. These costs include excess healthcare and travel costs, temporary incapacity compensation (pain and suffering), permanent functional and cosmetic incapacity, loss of income, and loss of life compensation. In order to receive compensation, the pharmaceutical injury claim must adhere to the terms and conditions of the Finnish Pharmaceutical Insurance Pool. For example, the pharmaceutical injury claim must be submitted within 3 years of the claimant becoming aware of the injury or no later than 10 years after discontinuing taking the pharmaceutical. Additionally, no compensation is rewarded if the pharmaceutical injury is regarded as tolerable in relation to the nature and severity of the illness being treated. 14 Health service use in Finland consists of visits to public, private, and occupational healthcare. Finland's publicly funded healthcare system is divided into three levels and consists of primary health centers, secondary central hospitals, and tertiary university hospitals. The healthcare system is organized by municipalities and divided into 21 larger hospital districts and five university hospitals. The private healthcare sector is much smaller and less frequently used. Additionally, employers are obliged to organize and fund occupational healthcare for employees. In this study, FQ-related AE costs were examined from a societal perspective as all AE-related costs, regardless of whom they fall on. FQ-related AE costs were divided into direct and indirect societal costs. Indirect costs of the employed claimants were estimated according to the friction cost approach. 20 Indirect costs accounted for the loss of productivity due to FQ-related tendon injuries during a friction period when a substituting employee was searched to replace the injured employee. The average friction period in hours was multiplied by the average hourly loss productivity estimate based on the Statistics Finland's average earnings and average social costs paid by the employer. 21 Average Finnish earnings were chosen due to lack of salary information in the data. The friction period was estimated from Statistics Finland's Employment service statistics 22 and following Tan et al., 23 an additional 4-week period was added to estimate the time employers require to decide to open a vacancy.
Due to elasticity between labor time and labor productivity, we applied a 0.8 elasticity measurement to indirect costs, as suggested by Koopmanschap et al. 20 Previous studies have shown hospitalization to be a costly form of health service use. 24,25 As there was a high occurrence of no hospital days in the dataset, a two-part model was built with R (version 3.6.2.) to predict hospitalization and the length of hospital stay. Additionally, two Generalized linear models (glm) were built to estimate the association of tendon ruptures and specifically bilateral ruptures with patient characteristics. Several subgroup analyses were done to various subsets of claimants, so that their age, sex, use of oral corticosteroids, the number of co-morbidities or the fluoroquinolone derivative used could be taken into account. All costs were converted to 2017 euro.

| Direct costs and health service use
The estimated average direct societal cost of a FQ-related tendon injury episode per claimant, including incapacity compensation, was 14,800€. Without incapacity compensation, a FQ-related tendon injury amounted to 8744€. The total direct costs associated with FQ-related tendon injuries among claimants were 2,146,057€.
Hospitalization was the costliest form of health service use. The claimants frequently visited a primary care facility (331 visits) and rehabilitation, such as physical and lymphatic drainage therapy (686 visits). The averages of health service use and costs are presented in conservatively (equinus cast for 3-9 weeks), while 33 (23%) claimants' tendon injuries were treated with surgery.

| Statistical analyses
Dichotomous and continuous independent variables were selected to determine the impact of known risk factors, AE severity and AE year on hospitalization. As approximately half of the claimants had no hospital days and the data exhibited over dispersion, a two-part model was built to combine the presence of excess zeros and the negative binomial distribution of counts in the data ( Table 2)  There were no substantial differences between males and females, different ages, treatment options, or AE years. However, claimants with co-morbidities had on average 11 hospital days in comparison to healthy claimants who had only 2 hospital days. Figure 3 provides an illustration of the impact of tendinitis, in addition to single and bilateral tendon rupture on the amount of hospital days and total direct costs.
Glm logit models predicting tendon ruptures among the claimants showed that advancing age slightly and taking oral corticosteroids more prominently increased the odds of suffering a tendon rupture (OR = 1.035 and OR = 2.55, respectively) ( Table 3). The impact of concurrent oral steroid use was more substantial among claimants with bilateral tendon ruptures (OR = 3.98, p = .00808) ( Table 4). The number of co-morbidities, and again, the prescribed FQs, AE year, and claimant gender were not statistically significant in these models.

| DISCUSS ION
Our study found that the 145 claimants with tendon injuries had an average of 14,800€ direct costs in addition to the indirect costs that were estimated to be 9077€ for employed claimants. Fifty-one percent were hospitalized, with an average duration of 21 hospital days.
Hospitalization was the costliest form of health service use amount-  long hospital stay in the patients who were compensated. In contrast to other serious AEs associated with FQs, such as cardiac toxicity, 34 tendon injuries are not a direct cause of mortality. However, research has shown that traumatic injuries are associated with significantly higher morbidity and mortality in elderly patients than in their younger counterparts. 35 The death of eight claimants during the compensation claim process affected the amount of injury compensations. It is a common practice for insurance companies to hold back the total compensation until the level of permanent disability has been evaluated. In such circumstances, the cost of a tendon injury episode will be underestimated.
Although the number of levofloxacin-based claims might give the impression that its tendency to cause tendon injuries exceeds that of other FQs, this is likely not the case. Forty-five percent of the claimants had previously been diagnosed with chronic lung diseases, which made them more susceptible to respiratory infections that were often treated with levofloxacin. Patients with chronic lung disease are frequently treated with inhaled glucocorticoids, as well as systemic glucocorticoids, which strongly increase the risk of FQ-associated tendon injuries. Moreover, it is possible that some of these patients had received fluoroquinolones on several occasions. The risk of tendon injuries appears to rise with an increasing number of fluoroquinolone prescriptions and with increasing cumulative dose 6 ; the accumulated risk has been estimated to be approximately 6% with each additional day exposed to FQs. 36 Thus, we believe that levofloxacin was overrepresented in the present study sample, because it was often used in high-risk patients.
EMA has recommended restricting FQ use in infections that are not severe, as a first-line treatment and to avoid prescribing to high-risk groups, such as those using corticosteroids. 3 With current recommendations, it may be possible to avoid some of the tendon injuries, but it is likely that in some cases there are no good alternatives to FQs. Moreover, some of the tendon injuries occur to patients who have no known risk factors, as also seen in our study, and, therefore, some of the cases remain unavoidable also with current knowledge.
Medical staff needs to stay wary of the risks in order to prevent and reduce fluoroquinolone-related tendon injuries, in addition to health service use and costs associated with them. However, due to the growing amount of antimicrobial-resistant bacteria, the rapid decline of effective antimicrobials, and alarmingly small number of new discoveries, AEs and resulting costs might have to be increasingly tolerated in the future, in order to combat bacteria. Additionally, decision makers need to be aware of the economic burden of AEs from a societal perspective as its extent is often underestimated.

ACK N OWLED G M ENTS
This research received no funding from any funding agency in the public, commercial, or not-for-profit sectors.

D I SCLOS U R E
The authors of this manuscript have no conflicts of interest to declare.

AUTH O R CO NTR I B UTI O N S
All persons who meet authorship criteria are listed as authors. All authors took part in conceptualizing the idea, designing the study, analyzing the data and writing the manuscript. All authors have seen and approved the final version of the manuscript.

E TH I C A L A PPROVA L
This is a retrospective observational study with register data, and neither an ethics approval nor patient consent was required.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy restrictions.