Study design

The participants were recruited from the AgeCoDe study, which is a multi-center longitudinal study, initially with a total of 3327 patients of at least 75 years of age in 2003/2004, that were recruited from GP patient registries in Germany with frequent follow-up periods (e.g., [14]). Medication was documented at every AgeCoDe follow-up assessment. The present CIM-TRIAD study was conducted at three German study sites (Bonn, Hamburg, and Leipzig).

Semi-structured interviews were chosen as a qualitative method to obtain comprehensive explanations for context and individual factors that contribute to chronically inappropriate medication. Dyads or triads of elderly patients, their attending GPs, and relevant others were interviewed. The patient perspective was examined in the present study. Patients were not explicitly informed about PIM characteristics of their drugs in advance.

Cases of chronic PIM intake and matched cases without PIM (non-PIM) were identified by BH (pharmacist and pharmacologist) and DP of the group pharmacoepidemiology (led by BH). Criteria for PIM cases included an age of 75 years or older and the intake of at least one drug from the PRISCUS list [3] in the last follow-up intervals available. If available, preference was given to patients who were taking drugs from the PRISCUS list continuously for as many follow-up intervals as possible, including baseline. A non-PIM group was interviewed as well, in order to identify contextual factors that might decrease the probability of PIM prescription in very elderly patients. Controls did not take any drug from the PRISCUS list in baseline and any follow-up intervals. Controls were matched to cases on age (if available, ± one year, otherwise ± maximum five years), gender and the number of prescription drugs during the AgeCoDe study (baseline and all follow-up intervals available). For the number of prescribed drugs, the control with the number closest to the number of prescribed drugs of the case was chosen.

Identified cases were contacted and asked to participate in the study after the procedure of the study was explained. If patients consented to participate in the study, they were also asked for their permission to contact their GP and another significant person, to ask them also for their study participation. We aimed to conduct interviews of full triads containing all three perspectives. We obtained 52 patient interviews (PIM: n = 27, non-PIM: n = 25), 52 GP interviews (PIM: n = 25, non-PIM: n = 22, others: 5; for results see [15]), and 48 interviews of significant others of the patients (PIM: n = 24, non-PIM: n = 24). Interviews of the triad were usually conducted separately. Characteristics of the patients with and without chronic use of PIM are given in Tables 1 and 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of the patients with chronic use of potentially inappropriate medication (PIM).

https://doi.org/10.1371/journal.pone.0202068.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of the patients without chronic use of potentially inappropriate medication (non-PIM).

https://doi.org/10.1371/journal.pone.0202068.t002

Development of the interview guideline

The interview guideline for the elderly patients was developed in Bonn (by KH and FJ). During the development of the interview guidelines, interdisciplinary exchange was obtained by the means of audioconferences and personal meetings. Audioconferences and personal meetings also served to assure homogenous implementation of the interview guideline and of the analysis strategies. Topics of the interviews were knowledge, awareness, risk monitoring and management, opinions, including opinions about the beliefs of the other individuals of the triad, and communication among the triads. Modifications of the interview guidelines after pre-test interviews were minor, only concerning the sequence of a few questions, and the rephrasing of some questions as open-ended questions.

Interviewer training

NP is an experienced qualitative interviewer. KH and AL were trained by NP in an interviewer workshop where two patients were interviewed using the guideline for patient interviews developed in Bonn. KH and AL received feedback about their interviewing techniques from the respective other researchers after the pilot interviews.

Data collection

Interviews were conducted between December 2014 and July 2015 by AL (Master of Public Health), KH, and NP (both postdoctoral psychologists). Patient interviews lasted between 24 and 121 minutes in Bonn, between 32 and 111 minutes in Hamburg, and between 16 and 54 minutes in Leipzig. Patients were interviewed face-to-face in the accommodations where they lived and the interviews were audiotaped.

Data analysis

Audiotapes were transcribed verbatim by a research assistant following designated transcription rules. All transcripts were checked for accuracy and corrected where required (e.g., language corrections). They were read repeatedly and important contents of all interviews were summarised in an abstract to get an overview over each triad and to facilitate the identification of important topics. The unabbreviated transcripts were analysed using qualitative content analysis [16] to condense the large amount of data to identify the main themes. The coding was conducted using MaxQDA version 11 (Verbi GmbH). At first, deductive categories were developed based on the literature and based on plausible investigator assumptions about the research objectives (e.g., communication between GP and patient, positive effects of PIM on quality of life, and barriers to deprescribe PIM drug). Inductive categories were added during the coding process when relevant themes emerged, or sub-categories of higher level deductive categories were identified. Categories were described in code memos. Contextual factors were grouped into factors that might increase the probability of chronic intake of PIM, derived from PIM interviews, and factors that might decrease the probability of chronic intake of PIM, derived from non-PIM interviews. The development of deductive categories was initially conducted by KH and FJ in close collaboration with NP and AL. The patient interviews were content analysed at the study site in Bonn (by KH) in close collaboration with the study site in Hamburg (NP) and Leipzig (AL). Inductive categories were developed by KH in close collaboration with FJ during the transcript reviews. The results of the qualitative content analysis were discussed during two interprofessional meetings of the CIM-TRIAD study group at different study sites (by KH, NP, AL, MS, FJ, SRH, BH, and DP) to secure intersubjective comprehensibility and credibility.

Ethical considerations

The study was approved by the local Ethics Committees (Ethics Committee of the Medical Faculty of the University of Bonn: July 14^th 2014, 169/14; Hamburg Medical Association: October 8^th 2014, MC-251/14; Faculty of Medicine, University of Leipzig: August 27^th 2014, 269-14-25082014). All participants gave a written, informed consent to participate in the study which included their explicit consent to contact and interview their GPs and associates.

Positive features of PIM

Positive features of the PIM perceived by the patients might contribute to the chronic use of that drug. For example, the perception of positive features might support the patient´s wish for a continuation of the prescription.

Maintaining characteristics of medication intake

There might be characteristics of the medication intake behaviour of patients that contribute to the maintenance of chronic prescription and of chronic use of PIM.

Barriers to deprescribe PIM

We identified different barriers that might prevent the deprescription of the PIM.

External actors supporting PIM intake

Besides the patient, physicians and relatives or other associates of the patient might influence the medication of the patient.

System-related factors

Some PIM are over-the-counter products.

Some patients reported that they bought their PIM (Doxylamin) as over-the-counter (OTC) product in pharmacies. Some of these patients did not inform their GPs about the intake of this drug on purpose.

P: “It was prescribed by my previous physician and they are available without prescription. And I´m also taking them since then.” I: “But you did not talk about that with doctor X so far?” P: “No, no, I will beware of doing that.” (P11, Doxylamin)

Beside the above-mentioned contextual factors that might increase the probability of chronic intake of PIM, we also found characteristics reported by patients without PIM intake that might indicate contextual factors that possibly decrease the probability of (chronic) intake of PIM (see Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Contextual factors that might decrease the probability of chronic intake of PIM (d = deductive category, i = inductive category).

https://doi.org/10.1371/journal.pone.0202068.t004

Health-related behaviour

Patient-GP-interaction

Medication-related attitudes and knowledge

Factors that might increase chronic PIM use

Positive features of the PIM motivated the patients to continue its use. Positive main and side effects, a good tolerance, a long-lasting PIM intake, and positive effects of PIM on quality of life were sub-categories of this contextual main factor. In case of a long-lasting intake, PIM might have been prescribed initially in younger ages and might not have been potentially inappropriate then due to the patient´s age. Good tolerance of PIM might contribute to the continued prescription, but adverse effects might develop later on due to aging processes. However, a PIM prescription at a younger age might also produce habituation, and the awareness of adverse effects might be reduced over time. Positive effects of PIM on the quality of life were predominantly reported for benzodiazepines and hypnotics in our study. Cook et al. [17] found that many older patients with long-term benzodiazepine use attach great importance to the positive effects of benzodiazepines, while negative side effects are neglected. “Means to cope with stress or anxiety” or the “ability to make them feel more like themselves” were identified in this study ([17] p. 1096), which are similar to positive effects on quality of life reported in the present study. Hence, patients that chronically use benzodiazepine PIM might be “active users” due to the appreciated positive drug features, whereas patients that use other PIM might be less active or rather “passive users”. In another qualitative study by Cook et al. [18], physicians minimised concerns regarding potential problems of benzodiazepine use and justified long-term use in the elderly (also due to a lack of signs of addiction). The patients´ resistance against benzodiazepine discontinuation was anticipated by the physicians who also did not consider it a central focus of their work to monitor or restrict benzodiazepine prescription in elderly patients [18]. Mah and Upshur [19] found more discordant patient and physician perceptions of long-term benzodiazepine use as patients rated these drugs more positive. Although benzodiazepines are PIM due to the high risk of adverse drug effects, it should be mentioned that ceasing a long-lasting usage in old age can also bear risks and requires medical management (e.g., [20]).

The interviews of patients with chronic PIM use often suggested that the patients were taking their drugs in a reasonable way or that the patients had a lack of knowledge regarding the PIM. Reports of patients on low dose benzodiazepine intake might be due to an underestimation of the real intake or due to low dose dependency, but it might also be a mode of drug intake that maintains chronic use, as these patients might be “reasonable users”. It seems plausible that GPs might be induced to prescribe a potentially problematic drug for rather careful patients, whereas GPs might not prescribe the same drug to noncompliant patients. Patients who do not know that they are taking a PIM or do not know its indication, might not associate potential side effects with the PIM and might therefore not have the opportunity to speak up for its cessation. A lack of knowledge regarding medication might also be associated with insecurity in elderly patients as Modig et al. [8] identified “deficient information”, “distrust”, and “lack of availability” as sub-categories of “insecurity with information” associated with the potentially modifiable factors “too short consultations, discontinuity, lack of availability for questions or opportunity to contact the physician if adverse effects were suspected” (p. 9).

Barriers to deprescribe the PIM drug were also reported. Some patients did not care about the side effects of their PIM. Other patients rejected alternative treatments or resisted the PIM cessation (e.g., due to dependency). In a study by Leydon et al. [21] on selective serotonin reuptake inhibitors (SSRI), “patient uncertainty about benefits of and continued need for medication”, “barriers to stopping”, and “the importance of GP´s role in facilitating cessation” were identified as main topics. As also reported by patients in our study, the fear of relapse and the fear of withdrawal symptoms were also reported in this study [21]. Verbeek-Heida and Mathot [22] also found that the fear of stopping the intake of SSRI exceeded the fear of its continuation.

External actors supported the PIM use and contributed to it as well. Some patients reported that although their GPs prescribed them PIM, they informed the patient about adverse drug effects and expressed their concerns. Other GPs seemed rather unconcerned about PIM and some prescribed it without personal contact over longer periods of time. Private prescriptions and cost acquisition by health insurance were both reported by (benzodiazepine) patients. This might be a strategy to regulate the intake of a drug (eventually encouraging a lower drug dosage, if the patient has to pay for it) or to conceal the prescription of certain drugs (e.g., from health insurances). Ageism expressed by the GP was also identified in another study as a sub-category that underlies the inappropriate use of medicines [12]. Some patients in our study also expressed statements of ageism, which might interfere with PIM cessation as a fatalistic outlook on the need to change drug schedules due to the own high age was implied. In rare cases, relatives supported the use of PIM.

System-related factors contributed to the chronic PIM use in some cases. For example, PIM drugs that can be bought OTC were consumed by some patients without informing the GP because the patients expected their disapproval. If drugs such as Doxylamin would be only available on prescription, some patients would possibly dispense with it. The problem that some PIM drugs can be bought OTC in the United States of America was also discussed by other authors [23]. Mental disorder, female gender, and number of long-term medication were patient factors that increased the probability of PIM prescription in another German study [24].

Factors that might decrease chronic PIM use

Patients without PIM use were also interviewed to generate hypotheses about “successful pathways” of the medication process. Characteristics of health-related behaviour, patient-GP-interaction, and medication-related attitudes and knowledge were identified as contextual factors. However, we want to point out that some of these characteristics were also expressed by patients with chronic PIM use. One potentially relevant health-related patient behaviour was a critical attitude towards medication in general combined with a conscientious usage of necessary drugs as these patients probably will not take dispensable medicine. Patients who adhere to GP instructions and give their GP a complete drug overview might reduce their risk for drug-drug-interactions. A patient-GP-interaction that is characterised by agreements on various aspects of medication might give patients the opportunity to report on adverse effects, but probably also requires a higher level of commitment of the patients. However, if practiced, agreements might reduce the risk of chronic PIM use, for example if (subtle) side or adverse drug effects exist.

Spinewine et al. concluded that “improvements targeted at the abilities of individuals, better doctor-patient and doctor-doctor relationships, and systems for transferring information between care settings” will lead to a more appropriate use of medication ([12] p. 1). These results are in good accordance with our results regarding “successful pathways” of drug prescribing as we also identified mutual agreements and close communication between patient and GP, which should stimulate treatment review, as possibly advantageous characteristics. Britten et al. [7] found that misunderstandings often occurred during the prescription of drugs and both patients and physicians contributed to this (i.e., “patient information unknown to doctor”, “doctor information unknown to patient”, “conflicting information given”, “disagreement about attribution of side effects”, “failure of communication about doctor´s decision”, “relationship factors”; p. 485). Although these results were not PIM-specific, it underlines the importance of close communication between patient and GP to prevent misunderstandings and to identify and eliminate side effects of the drugs.

Medication-related attitudes that were more frequently reported by patients without PIM use dealt with a personal reservation against hypnotics, analgetics, and psychotropic drugs, although a few patients without PIM use reported psychological stress for which alternative medical or non-medical treatments were used. Hence, although a documented mental disorder increased the risk for the prescription of PIM in another study [24], there are patients that favour alternative non-medical treatments for psychological distress or sleep disturbances. Former studies showed that psychotropic drugs such as psycholeptics/psychoanaleptics [4] and sedatives/hypnotics [24] account for a high percentage of PIM prescription. Hence, reservation against psychotropic drugs should reduce the possibility for PIM intake, at least regarding drugs of these agent groups.

Strengths and limitations of the study

Audioconferences, an interviewer training and personal meetings provided extensive and interdisciplinary exchange and confirmability during the development and implementation of the interview guidelines. The fact that modifications of the interview guidelines after pre-test interviews were mostly not required was interpreted as an indicator of credibility. The content analysis was conducted only by one coder for the patient perspective. Although this was a clear limitation of our study, we accomplished validity and confirmability of the coding by ensuring an intensive exchange on the coding system, the results and the presentation of the results during several audioconferences and personal meetings. Repeated reading and summarising of the full interviews also increased the confirmability of our results. Different PIM agents were included in our study and the duration of chronical PIM intake varied to some extent which should enhance the transferability of our results. Additionally, most of our participants were female which is in accordance with other studies showing a higher prevalence of PIM in women [4, 24]. Due to the design of our study, we did not include participants without associates and participants with severe cognitive impairment. Although these restrictions were necessary, they might have reduced the transferability of our results. The interview of triads contributed to the dependability of our results as cross-validation of reported contents automatically occurred to some extent. However, similarities between the group of PIM users and non-users, the dosage-dependent definition of some PIM, and the short- or long-term use of PIM, not regularly but as needed, might have diminished the dependability of our results.

The consideration of chronic PIM use enabled us to identify relatively stable contextual factors, instead of factors that might lead to short-term PIM use. We conducted a considerable number of patient interviews in our multi-perspective and multi-center qualitative study. The interview of triads and the inclusion of questions regarding quality of life were highly innovative. The integration of different professional perspectives on the obtained data was ensured by the different theoretical and educational backgrounds at the three study sites. The differing backgrounds of the patients lead to an enriched sample containing several pathways to chronic PIM use, or “successful” pathways without PIM use. As triads and different agent groups of PIM drugs were included, we planned a high number of interviews in advance. Therefore, saturation was not the only criterion for the number of interviews, but it was in good accordance with the rather high planned and presented number of participants. As there was an overlap to some extent between patients with chronic PIM use and those without, we aimed to find a reasonable compromise between higher-level conclusions and the potential danger of an oversimplification of the complex situation in real life contexts of the medication process in elderly patients. Although a lot of PIM cases were due to benzodiazepines, there was a wide range of different prescribed PIM agents in our sample overall.

Conclusion

The results of our study should be relevant for both physicians and developers of guidelines and educational programs. The avoidance of initial and long-lasting PIM use by restrictive prescription and the sensitisation of the patient for the risks of these drugs might be indicated (e.g., specific side effects or drug-drug interactions). It might also be helpful to inform patients more actively about adverse effects of prescribed PIM and to inquire adverse drug effects routinely. Test phases for newly prescribed drugs and a routine follow-up appointment to discuss potential side effects should be arranged. A frequent drug treatment review and close communication between GPs and patients, including mutual agreements on prescribed drugs, should be pursued. Advantages of alternative treatment approaches should be emphasised by GPs. Concerns regarding the cessation of PIM might be addressed and modified in GP consultations. Discontinuation and withdrawal in case of dependency might be supported and facilitated by GPs. Ageism might be identified and reduced in educational programs and GPs might assist patients and relatives in the modification of drug-related ageism. In addition to improved specific drug education, strengthening global health literacy in patients as a more complex approach to empower patients might also be promising to reduce chronic PIM use. Our findings also might be useful to generate hypotheses about chronic PIM use which could be tested statistically in quantitative study designs.