Association of polypharmacy with fall-related fractures in older Taiwanese people: age- and gender-specific analyses

Objective To elucidate the associations between polypharmacy and age- and gender-specific risks of admission for fall-related fractures. Design Nested case–control study. Setting This analysis was randomly selected from all elderly beneficiaries in 2007–2008, and represents some 30% of the whole older insurers using Taiwan's National Health Insurance Research Database. Participants We identified 5933 cases newly admitted for fall-related fractures during 2007–2008, and 29 665 random controls free from fracture. Primary and secondary outcome measures Polypharmacy was defined as the use of fall-related drugs of four or more categories of medications and prescribed related to fall within a 1-year period. Logistic regression models were employed to estimate the ORs and related 95% CIs. The interaction of polypharmacy with age and sex was assessed separately. Results Compared with those who consumed no category of medication, older people who consumed 1, 2, 3 and ≥4 categories of medications were all at significantly increased odds of developing fall-related fractures, with a significant dose–gradient pattern (β=0.7953; p for trend <0.0001). There were significant interactions between polypharmacy and age, but no significant interactions between polypharmacy and gender. The dose–gradient relationship between number of medications category and risk of fall-related fractures was more obvious in women than in men (β=0.1962 vs β=0.1873). Additionally, it was most evident in older people aged 75–84 years (β=0.2338). Conclusions This population-based study in Taiwan confirms the link between polypharmacy and increased risk of fall-related fractures in older people; and highlights that elderly women and older people aged 75–84 years will be the targeted participants for further prevention from fall-related fractures caused by polypharmacy.


GENERAL COMMENTS
Athough the similar result was reported by using the same database (Polypharmacy correlates with increased risk for hip fracture in the elderly: a population-based study. Medicine (Baltimore) 2010; 89: 295-9. You have cited it), this is a very straight-forward study to explore the association between polypharmacy and fall-related fractures in older people. Also the used methodology (nested casecontrol study) and the analysis seems adequate. The discussion comprises the essential topics. I therefore recommend to accept this manuscript for publication.

Taro Kojima
Department of Geriatric Medicine, The University of Tokyo REVIEW RETURNED 09-Feb-2014 GENERAL COMMENTS -This paper could be acceptable, pending the appropriate responses to the comments.
-The present study evaluated the association between polypharmacy and fall-related fractures in Taiwanese elderly people, and found that the number of medications is a risk of fall-related fractures in a stepwise manner.
There are a few comments about the manuscript.
Major points.
1. There are several drugs which are risks for fall. The authors counted the number of medications but did not assess the categories of drugs. Is the category of drugs available for this study? If so, is there any type of drugs which was associated with fallrelated fractures? Is there a difference in prescription between 65-84 years-old and 85 years or older?
2. Are there any specific comorbid conditions which were related to falls? (Components of CCI or other major diseases which are not included in components of CCI, e.g. hypertension, insomnia, and osteoarthritis.)

VERSION 1 -AUTHOR RESPONSE
Reviewer Name Kuan-Fu Liao Institution and Country Department of Internal Medicine, Taichung Tzu Chi General Hospital, Taichung, Taiwan Please state any competing interests or state 'None declared': None Declared Athough the similar result was reported by using the same database (Polypharmacy correlates with increased risk for hip fracture in the elderly: a population-based study. Medicine (Baltimore) 2010; 89: 295-9. You have cited it), this is a very straight-forward study to explore the association between polypharmacy and fall-related fractures in older people. Also the used methodology (nested casecontrol study) and the analysis seems adequate. The discussion comprises the essential topics. I therefore recommend to accept this manuscript for publication.
Authors' response: Thank you very much. We have revised the above information on competing interests of the revised manuscript (page 21, last line).

Reviewer Name Taro Kojima Institution and Country Department of Geriatric Medicine, The University of Tokyo
Please state any competing interests or state 'None declared': None declared Authors' response: Thank you very much. We have revised the above information on competing interests of the revised manuscript (page 21, last line).
Major points.
1. There are several drugs which are risks for fall. The authors counted the number of medications but did not assess the categories of drugs. Is the category of drugs available for this study? If so, is there any type of drugs which was associated with fall-related fractures? Is there a difference in prescription between 65-84 years-old and 85 years or older? Authors' response: Thank you very much for your kindness and valuable response. In respond to your recommendations, we have rechecked our data and observed additional findings. Based on these findings, we revised our content and highlighted in blue. The category of drugs is also available for this study. We have analyzed the specific category drugs in association with fall-related fractures (please see the suppl. table 1 below). We found that there is a difference in the category of medications prescribed in patients with different ages, including blood glucose lowering drugs and insulin (A10A, A10B), antiparkinson drugs (N04), and antipsychotics (N05A) (data do not show). Furthermore, there is a difference in prescription between 65-84 years-old and 85 years or older (please see the suppl. table 2 below).
In this study, we aimed to assess the cumulative effect of these categories of medications on the risk of fall-related fractures, rather than to evaluate the association of specific category of medication with risk of fall-related fracture. This is mainly because that the above mentioned drug categories have been found to be associated with fall-related fracture in a number of studies (French et al., 2005;Hartikainen et al., 2007). Therefore, we counted the number of medication category, and associated the number of categories with the risk of fall-related fracture. To make it clear, we have provided the information on drug categories in Table 1  2. Are there any specific comorbid conditions which were related to falls? (Components of CCI or other major diseases which are not included in components of CCI (e.g. hypertension, insomnia, and osteoarthritis.) Comments Using CCI for assessing disease burden of community-dwelling people is a little uncomfortable to this reviewer. It was originally established for assessing the prognosis of inpatients (J Chron Dis 40; [373][374][375][376][377][378][379][380][381][382][383]1987), who are in severer condition. But there are several papers assessing CCI for noninpatients, so it could be acceptable.