Evidence-based medicine is the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients. The practice of evidence-based medicine means integrating individual clinical expertise with the best available external clinical evidence from systematic research (Sackett et al. 1996). The problem in decision-making is that NNT can only be derived from large published randomized controlled trials (RCTs). These seldom include elderly patients, let alone patients with dementia. The International Conference on Harmonisation (ICH) guideline on geriatrics (E7) states that, for medicines intended for diseases characteristically associated with old age, 50% or more of the participants should be aged 65 and older. For medicines intended to treat diseases present in, but not unique to, older people, 100 or more participants aged 65 and older should be included. Studies show that only 1% of all participants in trials of these medicines are aged 70 years or older (Beers et al. 2014). Thus guidelines are often based on NNT for a selected population that is usually younger and healthier, without multimorbidity or polypharmacy, than the population for which the medicine is destined. For that reason, also the NNH in frail old patients can seldom be derived from these RCTs, which makes it difficult to make choices. Given the frail population in nursing homes, one can assume that the NNH is much lower and that drug efficacy is seldom better, than that reported in RCTs.

Many medicines are used off-label by nursing home residents, and this is especially the case for psychotropic drugs. Moreover, most psychopharmacological interventions for patients with dementia are not evidence based. Bearing this in mind, it is difficult for individual physicians to weigh the benefit and harm of medication in frail elderly patients.

Screening tools can help improve rational prescribing in nursing home populations. Implicit screening tools, such as the STRIP or the Medication Appropriateness Index, are judgment based, are patient specific, and consider the patient’s entire medication regimen. The use of these tools relies on expert professional judgment, requires good pharmacological knowledge, and is time consuming. Explicit screening tools, on the other hand, are far easier to use. Explicit screening tools contain lists of medicines that are potentially inappropriate for frail elderly patients. Examples of these explicit screening lists are the Beers criteria, the STOPP and START criteria, and the FORTA criteria. Most of these criteria are based on a combination of evaluation of the literature, Delphi panels with experts, and a final evaluation by a small group of experts. For all drugs included in these screening lists, the potential for causing harm outweighs the potential benefit. Drugs can be considered inappropriate based on their potential to cause harm regardless of morbidity (e.g., by causing sedation, anticholinergic properties, causing orthostatic hypotension, etc.) or only in combination with specific morbidity (e.g., decreased renal function, dementia). Some screening lists, such as the STOPP and START criteria and FORTA, also address the risk of underprescribing by listing drugs that can be considered beneficial for elderly patients. These screening lists are easy to use and can be applied with little clinical or pharmacological knowledge. A disadvantage is their rigidity. They should not be used as lists of forbidden drugs but as a warning. It is up to the prescriber to balance potential harms and benefits (using the WHO six step) and to decide whether or not the chosen drug should be prescribed. Multiple studies have shown that the use of screening lists reduces inappropriate prescribing and serious adverse drug reactions, even in nursing home populations. The use of explicit screening lists is not sufficient for a complete clinical medication review, which requires the use of implicit screening lists and patient involvement.

It is difficult to say which explicit screening method is the best for dementia patients in nursing homes. There is a large overlap between most methods, but they may differ with regard to the drugs included. The STOPP and START criteria were developed in Europe and the Beers criteria in the USA. In these regions, but even between countries within Europe, other drugs will be reimbursed or prescribed, so these methods should always be adjusted to the local situation.

The high prevalence of potential inappropriate medication use by nursing home residents shows that is not easy to stop medication once it has been started, for several reasons (Reeve et al. 2013a; Schuling et al. 2012; Kalogianis et al. 2016). The first barrier to stopping or withdrawing medication is the attitudes and beliefs of the physician. Some physicians believe there is appropriate evidence for prescribing medication, whereas others may be reluctant to change medication orders made by hospital specialists or feel they lack the education and experience to taper off or stop medications. Physicians may also be concerned that their patients may feel that they are “giving up on them” or “leading them to quicker deaths.” In addition, clinicians feel pressured to prescribe according to clinical guidelines, even though they are aware that such guidelines are rarely based on evidence from studies involving older populations and rarely address the issue of modifying clinical targets with advancing age or care goals. These considerations often lead to “therapeutic inertia” – “recognition of the problem, but failure to act.”

Patient’s beliefs and attitudes, or those of their caregivers, may also hinder the withdrawal of potentially inappropriate medication (see Fig.10.2). Studies involving community-dwelling elderly patients have shown that most elderly patients have a very limited knowledge of their medications. In majority, they state that they do believe that most medication they use is important since otherwise the doctor would not have prescribed it. On the other hand, studies with the patient’s attitude toward deprescribing tool show that the vast majority would reduce their medication use if advised to do so by their physician (Reeve et al. 2013b). Both patients and physicians may be influenced by cognitive bias. Status quo bias means that a preference for continuing with the status quo exists, especially if it has been the default for many years. Omission bias — being more willing to risk harm arising from inaction (continuing medication) than from action (stopping medication) — may also be a problem. For successful cessation of (potential) inappropriate medication, a stepwise approach is best (Best Practice Journal 2010):