Validation Study of Claims-based Definitions of Suspected Atypical Femoral Fractures Using Clinical Information

Transcription

1 薬剤疫学 Jpn J Pharmacoepidemiol, 1 1) June 16:1 〇 Original Article Validation Study of Claims-based Definitions of Suspected Atypical Femoral Fractures Using Clinical Information Shiro TANAKA 1, Hiroshi HAGINO, Akiko ISHIZUKA, Teruhiko MIYAZAKI, Takari YAMAMOTO 4, Takayuki HOSOI 5 Abstract Objective: Monitoring the incidence of atypical femoral fractures(affs)using medical claim databases is useful to assess the safety of long-term bisphosphonate exposure. Therefore, we aimed to validate the relationship between clinically-defined suspected AFFs and the candidate patients obtained from claims data at three hospitals in Japan. Design: A cross-sectional study involving three hospitals that perform bone fracture surgery and from which electronic medical record databases of diagses and procedures are available. Methods: Candidate patients were at the medical databases using two International Classification of Diseases, 1th Edition(ICD-1)codes(subtrochanteric fracture and fracture of shaft of femur)in the claims databases. These potential cases by claim-based definition were validated using clinically-confirmed information such as, the patient operation records, the discharge records, or radiographic imaging findings as suspected AFFs. Results: Among fracture cases in the hospitals, and 9 cases with subtrochanteric fracture and cases with femoral shaft fracture were identified based on the ICD-1 codes in the claims databases. Clinically confirmed subtrochanteric fracture had a sensitivity of 81.8%(95% CI: %), and a specificity of 1.%(95% CI: %). For femoral shaft fracture, the sensitivity was 8.1%(95% CI: %), and the specificity was 1.% (95% CI: %). In subgroup analyses, the sensitivities in patients over the age of 5 years with a single fracture site and with osteoporosis were relatively higher than in other subgroups. Conclusion: The claims-based definitions of suspected AFFs are accurate, indicating the value of pharmacoepidemiological studies using the National Receipt Database. (Jpn J Pharmacoepidemiol 16;1(1):1-19) Keywords: atypical femoral fracture, diagstic codes, administrative claims data analysis, validation Introduction Bisphosphonates, one of the standard pharmacological treatments for osteoporosis, increase bone mineral density and reduce the risk of incident fracture by inhibition of osteoclast- 1 Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Japan School of Health Science, Tottori University Faculty of Medicine, Japan Public Health Research Foundation, Japan 4 Medical Science, Lilly Research Laboratories Development Center of Excellence Japan, Eli Lilly Japan K.K., Japan 5 Kenkoin Clinic, Japan Address for correspondence: Shiro TANAKA, Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshida Koecho, Sakyo-ku, Kyoto , Japan tanaka.shiro.8n@kyoto-u.ac.jp

2 14 Original Article Table 1 Characteristics of the three hospitals Hospital Region Number of beds Annual number of outpatients Annual number of inpatients Annual number of any operations Mean hospitalization period(days) A Tokai area B Kanto area C Tokai area mediated bone resorption 1). In recent years, atypical femoral fractures(affs)have been reported in patients on long-term bisphosphonate therapy ) 6). AFFs has been defined as subtrochanteric or femoral shaft fractures after minimal or trauma. The revised case definition of AFFs was reported by a task force of the American Society of Bone and Mineral Research (ASBMR) 7). The trends in the incidence of subtrochanteric fracture and bisphosphonate use in the US elderly population from 1996 to 7 have been reported, and it was found that the rate of subtrochanteric hip fracture and bisphosphonate use increased in parallel 8). To determine whether prolonged bisphosphonate therapy is associated with an increased risk of subtrochanteric or femoral shaft fracture, a population-based, nested case-control study was conducted in Canada 9). And it was reported that treatment with a bisphosphonate for more than 5 years was associated with an increased risk of subtrochanteric or femoral shaft fractures among older women. Although several case reports of AFFs in patients on long-term bisphosphonate therapy have also been reported in Japan 1) 1), the precise incidence rate, the trend, and the causal relationship with bisphosphonates remains uncertain. Safety studies using claims databases is an appealing approach to the causal relationship between long-term bisphosphonate use and the incidence of AFFs. Since such analysis has been rarely performed in Japan, a validation study on the definition of AFFs using claim-based database is needed. We aimed to clarify the predictive accuracy to clinically-confirmed AFFs(as suspected AFFs)using claim-based definition. This study was planned based on the method previously reported in the population-based, nested case-control study and the appendix 9). Materials and Methods Ethics Statement The protocol of this study was approved by the Ethical Committee of Public Health Research Foundation and was conducted in accordance with the ethical guidelines for epidemiological research by Japanese Ministry of Health, Labor and Welfare. Because of retrospective nature of this study, written informed consent was waived in accordance with the Guidelines for Epidemiological Research Issued by the Ministry of Health, Labor and Welfare. Study Design and Data Source We collected anymized databases from three hospitals cross-sectionally. The hospitals have orthopedic departments that perform bone fracture surgery and their electronic medical record databases of the diagsis procedure combination (DPC)are available. The profiles of the institutes, including numbers of beds and patients, are shown in Table 1. Study Subjects and Claims-based Definitions The study subjects were patients who have International Classification of Diseases, 1th Edition(ICD-1)codes of the 6 types of fractures between April 1, 1 and October 1, 1 in the claims databases. Six types of fractures were consisted with fracture of thoracic vertebra (ICD-1: S.), multiple fractures of thoracic spine(s.1), fracture of sternum(s.), fracture of one rib(s.), multiple fractures of ribs(s.4), fracture of lumbar vertebra(s.),

3 Table Background of 55 patients with fractures in the claims databases 薬剤疫学 Jpn J Pharmacoepidemiol, 1 1) June 16:15 Item Total(N=55) N % fracture(n=9) N % fracture(n=) N % Fe Age, years(median[min-max]) B5 years Hospital A B C Hospitalization:Outpatients s s Fractures at multiple sites Co-morbidity:Osteoporosis % % 45.9% 8.% 5.8% 16.% 1.4% 5.6% 19.7%.6% 14.9% % % 1.%.%.%.% 77.8%.%.%.%.% % % 8.6% 1.% 4.4% 8.7% 6.9%.4%.% 8.7%.% Both outpatients and inpatient multiple fractures of lumbar spine and pelvis (S.7), fracture of other and unspecified parts of lumbar spine and pelvis(s.8), fracture of shaft of humerus(s4.), fracture of shoulder girdle, part unspecified(s4.9), fracture of other parts of forearm(s5.8), unspecified fracture of forearm(s5.9), fracture of head and neck of femur(s7.), pertrochanteric fracture(s7.1), fracture of lower end of femur(s7.4), multiple fractures of femur(s7.7), unspecified fracture of femur(s7.9), fracture of shaft of tibia(s8.), fracture of shaft of fibula(s8.4), fractures involving multiple regions of one upper limb (T.), fracture of spine, level unspecified (T8), sequelae of other fracture of thorax and pelvis(t91.), sequelae of fracture of arm (T9.1), sequelae of other fractures of lower limb (T9.). The six types of fracture were determined by an author(hh, a specialist of orthopedics)to secure a statistical detection power. Candidate patients with claim-based AFFs were identified based on two ICD-1 codes(subtrochanteric fracture(s7.)and fracture of shaft of femur(s7.))in the claims databases 9). The influence of sex, age, the presence or absence of accident and trauma, and osteoporosis and malignancy(icd-1: C to C97, D to D9)on the diagstic accuracy was also explored by subgroup analysis. Outcome The primary outcome was clinically confirmed suspected AFFs, defined by physician of each hospital with discharge records, operation records, or radiographic images selected from the electronic medical record databases. The data for the primary outcome was collected on February 1, 1 to October 1, 1. Statistical Analysis The positive predictive value(ppv), sensitivity, and the specificity and their exact 95% confidence interval(ci)of the claims-based candidate patients for identifying clinically confirmed(suspected)affs were calculated. Results The number of six types of fracture was 55 in the three hospitals. Major fracture types(over 1%)were fracture of lumbar vertebra(s.): 67 cases(.9%), fracture of one rib(s.): 88 cases(15.%), fracture of head and neck of femur(s7.): cases(11.8%). Nine cases with subtrochanteric fracture and cases with femoral shaft fracture were identified based on ICD codes in the claims databases(table ). The

4 16 Original Article Table Characteristics in the cases with subtrochanteric or femoral shaft fracture Fracture Age (Years) Gender Hospitalization Osteoporosis Multiple fracture Identified in claims 7 57 Outpatient and subtrochanteric 65 Outpatient Outpatient patientsʼ median age was 76 years with subtrochanteric fracture and 51 years with femoral shaft fracture. There were women(.%)in the subtrochanteric fracture group and 1 women(56.5%)in the femoral shaft fracture group. Seven(77.8%)patients with subtrochanteric fracture and 14(6.9%)patients with femoral shaft fracture were hospitalized. Osteoporosis was diagsed in (.%)patients with subtrochanteric fracture and (8.7%)patients with femoral shaft fracture. The number of cases clinically confirmed

5 薬剤疫学 Jpn J Pharmacoepidemiol, 1 1) June 16:17 Table 4 Diagstic accuracy of the claims-based definition of subtrochanteric fracture Definition Case Medical records Non-case Claims Case 9 Non-case 54 Sensitivity or specificity(95% CI) Abbreviation: CI, Confidence Interval Clinically-confirmed suspected AFFs 81.8(48. to 97.7) 1.(99.9 to 1.) Table 5 Diagstic accuracy of the claims-based definition of femoral shaft fracture Definition Case Medical records Non case Claims Case Non-case 5 57 Sensitivity or specificity(95% CI) Abbreviation: CI, Confidence Interval Clinically-confirmed suspected AFFs 8.1(6.1 to 9.9) 1.(99.9 to 1.) subtrochanteric fracture and femoral shaft fracture was 11 and 8, respectively(including both fractures: 1 case). Break down information in the cases with subtrochanteric fracture and femoral shaft fracture were show in Table. The sensitivity, specificity, and the 95% CI of subtrochanteric fracture in the claims databases compared to clinically confirmed subtrochanteric fracture are shown in Table 4. The sensitivity was 81.8%(95% CI: %), and the specificity was 1.%(95% CI: %). In femoral shaft fracture(table 5), the sensitivity was 8.1%(95% CI: %), and the specificity was 1.%(95% CI: %). Subgroup analyses divided by age, sex, fracture site, and co-morbidity in patients with subtrochanteric fracture and femoral shaft fracture are shown in Table 6 and Table 7, respectively. The specificities and PPV for both fractures in the claims databases were 1% to clinically-confirmed suspected AFFs. The sensitivities in patients over the age of 5 years with a single fracture site were relatively higher than in the other subgroups. Discussion This validation study showed that the predictive accuracy of suspected AFFs by clinicallydefinition and candidate fracture defined by two ICD-1 codes, subtrochanteric fracture(s7.) and fracture of shaft of femur(s7.), in the claims databases is high in relation to clinicallyconfirmed suspected AFFs. Epidemiological data indicates that subtrochanteric and femoral shaft fractures account for approximately 1% of all hip and femoral fractures in the elderly 1). Of these, around 15- % are atypical, and AFFs thus probably account for only about 4 to 14% of all fracture of the femure 14). In the present study, a total of 9 clinically-confirmed suspected AFFs were identified in the claims databases from the three hospitals. High sensitivities and specificities were

6 18 Original Article Table 6 Subgroup analysis of the diagstic accuracy for subtrochanteric fracture Item Category N N Sensitivity(95% CI) Specificity(95% CI) PPV(95% CI) All Age Sex Fracture site Co-morbidity <5 years B5 years Male Fe Single Multiple Osteoporosis (48. to 97.7).(. to 97.5) 9.(55.5 to 99.7) 1.(54.1 to 1.) 6.(14.7 to 94.7) 9.(55.5 to 99.7).(. to 97.5) 1.(15.8 to 1.) 66.7(1. to 1.) 1.(99.9 to 1.) 1.(99. to 1.) 1.(99.8 to 1.) 1.(99.7 to 1.) 1.(99.7 to 1.) 1.(99.8 to 1.) 1.(99. to 1.) 1.(99.4 to 1.) 1.(99. to 1.) 1.(66.4 to 1.) 1.(66.4 to 1.) 1.(54.1 to 1.) 1.(9. to 1.) 1.(66.4 to 1.) 1.(15.8 to 1.) 1.(15.8 to 1.) Abbreviation: CI, Confidence Interval; PPV, Positive Predictive Value Total, fracture Table 7 Subgroup analysis of the diagstic accuracy for femoral shaft fracture Item Category N N Sensitivity(95% CI) Specificity(95% CI) PPV(95% CI) All Age Sex Fracture site Co-morbidity <5 years B5 years Male Fe Single Multiple Osteoporosis (6.1 to 9.9) 7.(44.9 to 9.) 9.(64. to 99.8) 77.1(46. to 95.) 86.7(59.5 to 98.) 8.1(6.1 to 9.9) No cases 1.(15.8 to 1.) No cases 1.(99.9 to 1.) 1.(99. to 1.) 1.(99.8 to 1.) 1.(99.7 to 1.) 1.(99.7 to 1.) 1.(99.8 to 1.) 1.(99. to 1.) 1.(99.4 to 1.) 1.(99. to 1.) 1.(85. to 1.) 1.(71.5 to 1.) 1.(7.5 to 1.) 1.(69. to 1.) 1.(75. to 1.) 1.(85. to 1.) 1.(15.8 to 1.) Abbreviation: CI, Confidence Interval; PPV, Positive Predictive Value Total, fracture observed for both subtrochanteric and femoral shaft fractures. Moreover, the specificities and PPV divided by background characteristics were approximately 1%. The specificities and PPV were relatively higher than reported by other validation studies using ICD codes 15)16). and femoral shaft fractures may be strongly affected by the background factors such as age, sex, and co-morbidity affect sensitivity. Therefore, it was thought that an appropriate combination of these factors may increase the sensitivity. There are a few previous data about the incidence of subtrochanteric and femoral shaft fractures and the associations with bisphosphonates from a safety study using large-scale administrative databases in Japan. A major advantage of such a study is its large sample size which allows for in-depth investigations. Given the high sensitivities and specificities shown in this study, the National Receipt Database (NRDB)is expected to be useful to clarify the precise incidence rates of AFFs and the causal relationship with osteoporosis treatments including bisphosphonate in the Japanese population. There are three limitations of this study. First is the small number of clinically-confirmed suspected AFFs patients. However, it was reported that the incidence of actual AFFs was very low compare to the incidence of hip fracture 14). Second, we did t obtain the reason of fracture incidence(by traumatic or ntraumatic). Therefore traumatic fracture may be included. But higher sensitivity, specificity and PPV were also observed in the subgroup of age B5 years, the bias to this result may be

7 薬剤疫学 Jpn J Pharmacoepidemiol, 1 1 June 16:19 relatively small. Third, suspected AFFs in this study do t exactly match to definition of AFFs by ASBMR task force. Additional study may be needed to clarify the relationship between clinically-defined AFFs and gold standards of AFFs defined by the methods of ASBMR. In conclusion, the claims-based definitions of suspected AFFs is accurate, indicating the value of pharmacoepidemiological studies using the NRDB. Ackwledgement The authors would like to thank Dr. Hiroshi Watanabe for suggestion to this study and also thank to Convergence CT Japan K.K. for the data collection from the primary clinical institute and for discussion. Conflict of Interest This study was conducted by funding support of Eli Lilly Japan. TY is an employee of Eli Lilly Japan K.K. The other authors declare that they have conflicts of interest with respect to this article. Contributors Guarantor: TH. Study concept and design: TH, HH, ST, TY. Statistical analysis: ST. Drafting of the manuscript: ST. Data interpretation and critical revision of the manuscript for important intellectual content, writing of the report, and approval of the final version: TH, HH, ST, AI, TM, TY. References 1 Bilezikian JP. Efficacy of bisphosphonates in reducing fracture risk in postmepausal osteoporosis. Am J Med 9; 1 Suppl : S14-1. Goh SK, Yang KY, Koh JS, et al. insufficiency fractures in patients on alendronate therapy: a caution. J Bone Joint Surg Br 7; 89 : Lenart BA, Lorich DG, Lane JM. Atypical fractures of the femoral diaphysis in postmepausal women taking alendronate. N Engl J Med 8; 58 1 : Neviaser AS, Lane JM, Lenart BA, Edobor-Osula F, Lorich DG. Low-energy femoral shaft fractures associated with alendronate use. J Orthop Trauma 8; 5 : Kwek EB, Goh SK, Koh JS, Png MA, Howe TS. An emerging pattern of subtrochanteric stress fractures: a long-term complication of alendronate therapy? Injury 8; 9 : Abrahamsen B, Eiken P, Eastell R. and diaphyseal femur fractures in patients treated with alendronate: a register-based national cohort study. J Bone Miner Res 9; 4 6 : Shane E, Burr D, Abrahamsen B, et al. Atypical subtrochanteric and diaphyseal femoral fractures: second report of a task force of the American Society for Bone and Mineral Research. J Bone Miner Res 14; 9 1 : Wang Z, Bhattacharyya T. Trends in incidence of subtrochanteric fragility fractures and bisphosphonate use among the US elderly, J Bone Miner Res 11; 6 : Park-Wyllie LY, Mamdani MM, Juurlink DN, et al. Bisphosphonate use and the risk of subtrochanteric or femoral shaft fractures in older women. JAMA 11; 5 8 : Kaji Y, Kabata T, Watanabe K, Tsuchiya H. Histological finding of atypical subtrochanteric fracture after long-term alendronate therapy. J Orthop Sci 1; 17 : Ishizuna K, Ota D, Fukuuchi A, et al. A case of femoral diaphyseal fracture after long-term treatment with zoledronic acid. Breast 15; 1 : Sasaki S, Miyakoshi N, Hongo M, Kasukawa Y, Shimada Y. Low-energy diaphyseal femoral fractures associated with bisphosphonate use and severe curved femur: a case series. J Bone Miner Metab 1; 5 : Nieves JW, Bilezikian JP, Lane JM, et al. Fragility fractures of the hip and femur: incidence and patient characteristics. Osteoporos Int 1; 1 : Giusti A, Hamdy NA, Papapoulos SE. Atypical fractures of the femur and bisphosphonate therapy: a systematic review of case/case series studies. Bone 1; 47 : de Achaval S, Feudtner C, Palla S, Suarez-Almazor ME. Validation of ICD-9-CM codes for identification of acetamiphen-related emergency department visits in a large pediatric hospital. BMC Health Serv Res 1; 1: Goldberg DS, Lewis JD, Halpern SD, Weiner MG, Lo Re V rd. Validation of a coding algorithm to identify patients with hepatocellular carcima in an administrative database. Pharmacoepidemiol Drug Saf 1; 1 : 1-7.! Manuscript received August, 15; % " & " revised May 6, 16; & $ accepted May, 16 (