Download Malmö November 19th, 2013 Dear Editor, Thank you for providing

Malmö November 19th, 2013
Dear Editor,
Thank you for providing us an opportunity to revise our manuscript.
The reviewers’ and your comments were very useful and our manuscript is now improved. All
changes suggested by you and the reviewers have been incorporated in the manuscript.
Our responses to the comments are listed below in blue text, following the questions.
Reviewer: Allen Huang
Major revisions:
1. Background section needs to be improved. The authors muddle references from different
countries into the text. I think the message is to highlight the problem of drug-related falls in
Sweden. They should clearly state this and that the global literature already shows varying
evidence. Ref 5 (Canada), Ref 6 (Finland), Ref 7 (US).
Answer: We deleted Ref 5 and chose to refer to Sjöberg et al 2010 showing extensive
treatment with FRIDs in Swedish hip fracture patients (Ref 7).
2. Background: The authors should clearly state the objectives of the study.
Answer: We wanted that focus should be on FRIDs and ODs, therefore the objectives of the
study have been clarified both in abstract and in the background part.
3. Methods: Please supply demographic information so comparisons of the study can be done.
What is the population of the province of Skåne? What is the age distribution? Denominator
of people older than 75 years? What is the multi-dose drug dispensing system? How does it
work? Why was it chosen to be studied? Why were both people living in the community and
those in nursing homes included? These populations have different falls risks. Please include
the PHASE-20 question items, perhaps in an Appendix. Over what time period were the data
collected?
Answer: We added a demographic description in the methods part and explained closer the
MDD system. We also added an explanation why the MDD system was chosen (see last
paragraph in the Background part). The PHASE-20 formulary is a validated Swedish formulary
(in Swedish) and is found in the reference list. We can also upload it as an additional file.
We also added information about the time period for the data collection.
4. Methods: the entire description of the randomized control trial background does not seen
relevant, since the data used was taken before allocation into the cohorts.
Answer: We removed this part from the Methods section.
5. Data collection: Please define “mild” falls and “severe” falls. Were the number of FRIDs
unique by ATC? If more than one pill size was used to build a desired dose did it count as 2
drugs? (eg. 20 mg tab + 10 mg tab).
Answer: We added the definition of mild and severe falls in the methods section. We also
added this phrase: “Every drug was counted as one with its unique ATC code regardless of
the dosage or number of pills for the individual patient.”
6. Results: Please include the total number of studies in the main text. Although n=369 is
included in Table 2 it makes it easier to follow if the number is explicitly indicated. A majority
lived in nursing homes- it is also better to indicate the percentage (76%) alongside for clarity.
The results section is muddled with data coming from different sources. Patients’ symptoms
are from the PHASE-20 questionnaire. Did all 369 patients have complete data from all
sources?
Answer: We completed the first phrase in the results part with the explicit numbers. We also
divided the description of results following the classification in mild and severe falls for a
better understanding.
7. Discussion: statistical methods chosen do not generate odds ratios, which would be more
comparable to other studies. Suggest the authors repeat the analysis separating out
community-dwelling patients from those living in nursing homes. Again without describing
the denominator population in the methods it is difficult to assess the generalizability of the
results. Authors report that antidepressants and anxiolytics were the most frequently used
FRIDs but there is no table showing the frequency analysis of all FRIDs in the study.
Answer: Twenty-four percent of the studied patients lived in the community and we believe
that the number has not enough power to analyse the population in the two separate
groups. We added a Table (Table 5) with the frequency analysis of FRIDs and ODs according
to the NBHW list.
Minor revisions: Background, para 3 sentence 5: “medications are a SIGNIFICANT risk factor”
instead of VITAL. Sentence 6: please fix the English.
Answer: We changed the term “vital” to “significant” as suggested and reformulated
sentence 6.
Reviewer: Michael Denkinger
Abstract:

Minor: The conclusion cannot be made: “would improve the quality of drug treatment in
primary care”. Only with regard to the outcome studied, but not in general. Patients could
suffer from severe depression or psychotic pathology and might experience better QOL with
the prescribed FRIDs.
Answer: We reformulated the cited statement to “might improve the quality of drug
treatment”, see abstract.
Background:

Minor: The statement “no consensus definition of falls” cannot be made. The presented
consensus is indeed one and should be cited from the Profane group: Lamb, Jorstad-Stein EC,
Hauer K, Becker C. On behalf of the Prevention of falls Network Europe and Outcomes
Consensus Group. J Am Geriats Soc 2005; 53:1618-2 and not from Merck Manual
Answer: We changed the phrase according with the comment and added the suggested
reference.

Minor: ODs not explained (only in abstract) and from the intro it is not clear why the
differentiation between ODs is made in the analysis as they are just another FRID?
Answer: We chose to assess data for FRIDs and ODs as classified of NBHW and this is now
clarified in the last paragraph from the Background section.
Methods:

Major: I do not see comorbidity or some sort of indication/disease as an independent
parameter that has been controlled for? This is the big issue with medication data: is it the
disease (e.g. depression) or the FRID that causes the fall?
Answer: We did not collect data on patients’ diagnosis and we added this comment in the
Discussion part as a major limitation of the study as followed: “A major limitation of the
study is also the lack of geriatric assessment as cognitive impairment, comorbidity and
disability which is necessary to get a deeper understanding of the causes of falls.”
Results:

Major: Any geriatric assessment? There seems to be rather high comorbidity, frailty and
disability with respect to the setting and the high number of falls but we don’t really know?
ADLs? Cognition? Gait speed or timed up and go?
Answer: See previous comment. We discussed it as a major limitation of the study.
Presentation issues:

Minor: Fig 3 should be following Table 1 as it gets into detail f what kind of medication and its
prevalence
Answer: We initially chose to present data in Fig 3 in the end of the Results section because
distribution of drug types is a secondary outcome parameter. We removed Fig 3 because we
believe that the same results are shown in Table 5.

Minor: Fig 1-2 should rather precede Table 3 or be omitted because- although less impressive-
no of FRIDs has been shown to be associated with falls in Table 3.
Answer: We agree in this matter and have removed Figure 1 and 2 from the paper.
General aspects:

Major: Wouldn´t it be interesting to test what kind of FRID really drives the associations? You
have high number of falls and FRIDs; too little power? At least in a bivariate fashion
presented as a figure (instead of Fig 1-2)
Answer: We believe that we have too little power to assess the association between falls and
separate type of FRIDs. However, due to your suggestion we chose to add a table (Table 5)
showing the frequency of different ATC groups among FRIDs and ODs.
Major: What about dosage: Zolpidem 3.5 mg is different from 7 mg, the same might account
for other drugs, especially ODs which again might explain the missing associations: i.e.
Metoprolol 50 mg versus 200 mg defined daily dosage? You could rate the drug according to
high, medium, low dosage to get a deeper understanding of your total findings.
Answer: We did not collect data about drug dosage and this is indeed a limitation of the
study. We chose to add this information as a limitation of the study according to reviewer´s
comment.
Discussion:

Minor: Info on not merging FRIDs and ODs should appear in the methods and background.
Answer: This information has been added both in the Background section (last sentence) and
in the Methods part.

Major: Here the main problem of the study is nicely discussed: no information on comorbidity.
Is there no possibility to get hold on that? At least some geriatric assessment? Otherwise it
could also only be some kind of frailty syndrome associated with the number of FRIDs that
drive the association.
Answer: We did not collect data on patients’ diagnosis and we added this comment in the
Discussion part as a major limitation of the study as followed: “A major limitation of the
study is also the lack of geriatric assessment as cognitive impairment, comorbidity and
disability which is necessary to get a deeper understanding of the causes of falls.”
Limitations:

Major: I do not see why a fall risk assessment tool would have improved interpretation of the
data. It´s the other missing factors discussed above: no data on
comorbidity/function/frailty/activity (should be in the limitations).

Answer: We added this comment in the Discussion part, see previous answer.

Major: Further limitations: cross-sectional, falls only addressed by patients report, no
prospective fall calendar.

Answer: We completed the Limitations part with a discussion about study design as
suggested by the reviewer.
Conclusion

Minor: See my comment above (overall and abstract).
We hope our manuscript is now acceptable for publication.
Sincerely,
Veronica Milos, MD, GP, PhD student
Faculty of Medicine, Lund University
Department of Clinical Sciences Malmö
Clinical Research Centre (CRC), building 28, floor 11
Jan Waldenströms gata 35, Skåne University Hospital
205 02 Malmö, Sweden
Mail: [email protected]
Mobile +46-733565679
Fax +46-42-4060857