First of all, we would sincerely like to thank Bocti and colleagues and Coyle-Gilchrist and colleagues for their interest and for providing such fruitful comments about our article.

Most of them, directly or indirectly, focus on the possibly non-causal nature of the association we found between benzodiazepines (BZDs) and dementia. To make this point clear, we never concluded that this association was causal. We designed our study in order to attempt to go further than previously published studies by minimizing possible confounding and reverse causation biases. Among others, this is the reason why our analyses considered only actual incident and not prevalent users at the expense of a drastic reduction in sample size.

Both rapid responses pointed out that symptoms that are possible indications for BZD prescription (anxiety, sleep disorders and depression) may precede dementia for years, possibly 10 or more, whereas the median duration of follow-up in our study was 6.2 years. We obviously agree but would like to add the following: (i) despite being about 7 years on average, the duration of follow-up was longer in a significant number of subjects, which made it possible to show that the excess risk in exposed subjects was mainly observed after 7 or 8 years, with a stronger trend after 10 years; (ii) in the case-control analysis, a significant association was found for the most distant exposures (starting 8 years or more before dementia) and not for recent ones; (iii) studies conducted on symptoms observed during the prodromal phase of dementia (1) clearly show that their frequency increases when approaching dementia onset. In that respect, if BZD prescription was mainly justified by such prodromes, the association would be expected to be stronger for treatments started a few years before dementia and not the reverse, as shown by our results.

1. Amieva H, Le Goff M, Millet X, Orgogozo JM, Peres K, et al. (2008) Prodromal Alz-heimer's disease: successive emergence of the clinical symptoms. Annals of neurology 64: 492-498.

Competing interests: None declared

In their systematic prospective study, Drs. Sophie Billioti de Gage and coauthors investigated an association between a use of benzodiazepines (BZDs) and incident dementia, to find that new BZDs use was associated with an elevated risk of dementia (multivariate adjusted hazard ratio 1.60, 95% confidence interval 1.08-2.38).(1) The authors prudently advised against unthoughtful BZDs usage in the elderly population, but caution is necessary in interpreting their crucial findings; namely a potential causality of BZDs for dementia versus a mere association.
First, no distinction was made among Alzheimer type dementia versus vascular or other types of dementia (probably due to a limited sample size in each category of dementia and a resultant low power), and as such a possible contribution of BZDs in each subtype of dementia remains unknown.
Second, no information was available on other important variables known to affect dementia such as socioeconomic status, physical activity, cigarette smoking as well as family history of dementing disorders and genotypes including apolipoproteins.(2)
Third and most importantly, if one assumes a causality rather than just an association, there would be a dose-effect relationship and may be a certain threshold of BZDs to cause dementia. Here, the cumulative dose is driven by the duration multiplied by daily dose. The first question is, the longer the exposure to BZDs the worse the risk of being demented? This point is not clear in Fig. 4 because of unreliable estimates due to attritions. The second question is, the higher the dose of BZDs the worse the cognitive outlook? This point is entirely unknown since no information on the daily dose was available, which leaves clinicians at a loss about which BZDs doses might not be justified (or whether even pro re nata dosage should be discouraged). Further, there was no information on adherence but it is too optimistic to assume a perfect adherence in an elderly population.
Fourth, apart from a well-known adverse effect on cognition, mechanisms of BZDs to lead to dementia remain to be elucidated. It is important to note that cognitive adversities should improve upon discontinuation from BZDs but this study found a statistically elevated risk among remote users only.
Finally, the focus was on BZDs and dementia in this study but they may not be a sole offender. In fact, many other psychotropics are increasingly (and probably too widely) utilised in the real-world in an absence of good evidence.(3) This precious database might be reanalysed by simply replacing the use of “benzodiazepines” with other often utilised psychotorpics such as “antipsychotics”, “antidepressants” or “mood stabilisers”. In so doing, a relative magnitude in potential risks for each class of psychotropic medications with respect to an association with dementia can be critically appreciated.
Nonetheless, the authors remain conservative enough to use the term “association” in the text and their conclusion that caution is necessary against unthoughtful BZDs use appears to be reasonable. Altogether, the use of medications (BZDs in this instance) should be kept to a minimum and simple to still be effective, which reflects the very basic of medicine.(4)

References:
1. Billioti de Gage S, Bégaud B, Bazin F, Verdoux H, Dartigues JF, Pérès K, et al. Benzodiazepine use and risk of dementia: prospective population based study. BMJ. 2012 Sep 27;345:e6231. doi: 10.1136/bmj.e6231.
2. Stephan BC, Kurth T, Matthews FE, Brayne C, Dufouil C. Dementia risk prediction in the population: are screening models accurate? Nat Rev Neurol. 2010 Jun;6(6):318-26.
3. Guthrie B, Clark SA, McCowan C. The burden of psychotropic drug prescribingin people with dementia: a populationdatabase study. Age Ageing. 2010 Sep;39(5):637-42.
4. Wolf MS, Curtis LM, Waite K, Bailey SC, Hedlund LA, Davis TC, et al. Helping patients simplify and safely use complex prescription regimens. Arch Intern Med. 2011 Feb 28;171(4):300-5.

Competing interests: None declared

We read with great interest the study by Billoti-De Gage et al. The authors report an increased risk of dementia associated with the use of benzodiazepines in PAQUID, a population-based longitudinal study. In our opinion the biological plausibility of a causal association between benzodiazepine use and a relentlessly progressive degenerative disorder such as Alzheimer's Disease is not supported by available evidence. In contrast, short-term effects of benzodiazepines on cognition are well documented (Tannebaum et al. Drugs Aging 2012; 29: 639-658). In our clinical practice, many patients presenting with cognitive impairment will improve by decreasing dosage of the drug, especially with long-acting benzodiazepines with active metabolites, such as flurazepam. This indicates a probable effect of drug accumulation, but we cannot conclude that this effect is an argument that benzodiazepines cause dementia.

More fundamentally, we believe there is a limit of the study that is not sufficiently emphasized in the discussion: though the authors claim the study has a follow-up period long enough to overcome the hypothesis of reverse causation, we believe this is not the case. In the main analysis the delay between exposure and outcome (median 6.2 years) is too short to support the conclusion of an increased risk attributable to the use of benzodiazepines per se. Indeed, there is evidence from several longitudinal studies, including PAQUID (Amieva et al. Ann Neurol 2008;64:492–498), that many subtle changes are taking place at the psychological and cognitive levels in the period of approximately 10 years before dementia becomes clinically evident (Sliwinski et al. Psychology and Aging 2003:18;658–671, Reisberg et al. Alzheimer’s & Dementia 2008:4;S98–S108).

This period is very likely characterized by subtle cognitive changes that are not always detected by standardized tests and scales. The psychological symptoms are non-specific and hard to recognize as part of the prodromal phase of dementia, and could include poor subjective sleep quality and anxiety, symptoms that could in turn be the reason for benzodiazepine use. Since the period of observation for the main outcome of this study is well within published duration for the putative prodromal period (5 to 14 years), it is unlikely that we can conclude that benzodiazepines are causally related to dementia. In conclusion, the more parsimonious explanation appears much more likely: these drugs are prescribed as part of the treatment of the early symptoms caused by the dementing process, but prior to the clinical diagnosis of dementia.

Competing interests: None declared

As a practising psychiatrist for 35 years, I find that this is a flawed study.The type of treatment I use is psychotherapy with or without medications.

The article makes conclusions on a class of medications that have different effects, half life and indications. I use Alprazolam and occasionally Lorezepam. From psychotherapy, it is clear that these medications I use are safe and don't cause dementia.

Too many patients are prescribed medications by a family doctor who do not understand the emotional issues causing their patients' anxiety or depression. There is no monitoring of their medications, no understanding of the specific benzodiazepine they prescribe and total lack of interest in abuse of alcohol that is an epidemic in adolescent patients and the elderly.

Patients are over medicated,taking too many medications for many ailments that are symptoms of depression and anxiety.

American doctors no longer talk to patients. They are too involved in compliance matters and insurance problems. They base their prescribing on the drug representative of the pharamceutical industry who now finance their drug research.

Medicine is an art not a science. Science is a method of trying to make some order of the complexity of life. We are just scratching the surface of understanding.

Much is distorted by business model where the powerful corporations as in healthcare distort information to increase their profit. Money and the profit motive do not have a conscience.

Warren S. Kriedman, M.D.
www.wkriedmanMD.com

Competing interests: None declared