Introduction

Between 2000 and 2050, the proportion of the global population that is 60 or older will double from 11% to 22%. In absolute numbers, this group will rise from 605 million people to two billion, representing an important demographic transformation [1]. Aging is a gradual and inevitable process that manifests itself primarily in morphological and physiological changes, making this transformation even more relevant. The International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) regards older adults as a special population “which differs of younger adults in terms of their comorbidity, polypharmacy and pharmacokinetics, as well as their increased vulnerability to adverse drug reactions” [2].

Multimorbidity leads to prescribing more medications, which favors polypharmacy and in turn raises the risk of potentially inappropriate prescribing (PIP) and adverse health events. Excess polypharmacy (≥ 10 medications) is considered a 5-year mortality predictor [3,4,5]. Inappropriate medications definition comprises those drugs for which the risk of suffering a resultant adverse event is higher than the clinical benefit. Inappropriate medications also include frequent use of medications, use of medications for longer than indicated, use of medications with a high-risk of drug–drug or drug–disease interactions, duplication of medications or prescribing drugs from the same therapeutic group, and not prescribing drugs that clinically should be prescribed [6,7,8].

Several studies have documented high rates of inappropriate prescribing in older adults, such as prolonged use of psycholeptic medications and/or nonsteroidal anti-inflammatory drugs (NSAIDs), which can reach 40% [6, 9, 10]. In addition, recent data have shown that almost half of older adults may be at risk of polypharmacy, PIP, and drug–drug interactions [11]. However, it has also been reported that polypharmacy does not necessarily increase the risk of PIP and that strategies directed at reducing polypharmacy to decrease PIP have not been successful [12].

The identification of factors associated with PIP is relevant for designing strategies focused on ensuring patients’ access to potentially beneficial treatments, and protecting patients from inappropriate prescribing and its potential harm [13]. To contribute to improving knowledge in this field for Mexico, the goal of this study was to analyze PIP, its prevalence and patient conditions associated with this phenomenon, in a cohort of older adults receiving outpatient care.

Methods

A secondary analysis was conducted using the database of the study “Cohort of obesity, sarcopenia and frailty of older Mexican adults” (COSFOMA) (Cohorte de obesidad, sarcopenia y fragilidad de adultos mayores Mexicanos) [14]. We analyzed data from 1252 adults aged 60 and older, who attended one of the 48 primary care centers known as Family Medicine Units (Unidades de Medicina Familiar [UMFs]) in Mexico City, during 2014. These centers are part of the Mexican Social Security Institute (Instituto Mexicano del Seguro Social [IMSS]), a public institution that provides medical care to over half of the country’s population.

Variables

Based on the information contained in the original database through patients’ self-reporting, the following data were extracted to integrate the variables for this study.

Sociodemographic data: sex, age, partnership status, and level of education (basic education ≤ 10 years, and higher education > 10 years).

Medications

Drugs regularly used for the treatment of chronic diseases. Medications were registered by the name of their active ingredient and then classified according to the anatomical therapeutic chemical (ATC) classification system [15].

Polypharmacy

Defined as the use of five or more medications [16].

Potentially inappropriate prescribing

From the list of medications, identification of drugs included in the 2015 Beers Criteria [17], and the application of the criteria for potential drug–disease or drug–syndrome interactions, based on the clinical information of each patient.

Chronic diseases

Hypertension, heart failure, type 2 diabetes, hypothyroidism, rheumatoid arthritis, psychosis, peptic ulcers, neoplasms, chronic obstructive pulmonary disease (COPD), cerebrovascular disease, Parkinson’s disease, neuritis, peripheral vascular disease, chronic kidney failure, anemia, and hepatopathy. Criteria for major depression were established using the center for epidemiologic studies depression scale-revised (CESD-R) [18].

Multimorbidity

Defined as the presence of two or more chronic diseases.

Other clinical conditions

Functional dependence, evaluated using the Lawton–Brody instrumental activities of daily living scale [19]; cognitive impairment determined by a mini-mental examination score of ≤ 23 [20]; and frailty, classified into three categories using the Fried frailty criteria (frail, pre-frail, and robust) [21].

Statistical analysis

A descriptive analysis of the sociodemographic variables, chronic diseases, medications, PIP and other clinical conditions was performed. Bivariate analysis was conducted by calculating the crude odds ratio (OR) and 95% confidence intervals (95% CI) to identify the association between the independent variables and PIP. A correlation matrix was used to detect collinearity among independent variables. Then, three multivariate binary logistic regression analyses with the significant variables from the bivariate analysis were conducted, employing the strategy of incremental variable inclusion into each model (forward step). First, sociodemographic variables were introduced, along with polypharmacy, multimorbidity and other clinical conditions (functional dependence, cognitive deterioration, and frailty) (Model 1). Next, the multimorbidity variable was replaced by the individual diseases that were statistically significant in the crude analysis (Model 2). Finally, the same variables from Model 2 were introduced, and medications that had significance in the crude analysis were incorporated, and the polypharmacy variable was removed (Model 3). A p value < 0.05 was considered statistically significant. The analyses were conducted using version 11 of the statistical software STATA (StataCorp, Texas).

Results

Women comprised 59.9% of the 1252 cases analyzed. The average age of the total sample was 68.4 ± 7.1. According to the Beers Criteria, 524 (41.8%) participants had at least one PIP. The subgroup of patients with PIP was characterized by higher proportion of women; higher age; more patients living alone; lower educational levels; higher proportion of patients with functional dependence, cognitive deterioration and frailty; and higher number of cases with multimorbidity and polypharmacy (Table 1).

Table 1 Characteristics of study population
Full size table

Patients with PIP had higher rates of depression (35.5%), hypertension (24.6%) and type 2 diabetes (18.9%). These patients more frequently reported taking medications across all of the groups of drugs analyzed in this study, especially anti-inflammatory-antirheumatic and analgesic drugs (75.4%), antihypertensive drugs (50.6%), and hypoglycemic agents (29.6%) (Table 2). Of the medications to be avoided according to the Beers Criteria, nonsteroidal anti-inflammatory drugs (NSAIDs) and glibenclamide were the most frequently prescribed. The main drug–disease interaction was the use of clonazepam in patients with cognitive impairment (3.2%), and the principal drug–drug interaction was the use of warfarin together with NSAIDs (Table 3). The crude odds ratio for all variables showed significant association with PIP, except for rheumatoid arthritis or neoplasms (Table 4). Multivariate models with adjusted associations resulted in the following: in Model 1, polypharmacy (OR 6.88, CI 95% 4.38–10.81, p <0.001) and the clinical condition of frailty (OR 1.60, CI 95% 1.03–2.47, p = 0.037) were statistically associated with PIP. In Model 2, only the polypharmacy variable (OR 6.81, CI 95% 4.33–10.75, p <0.001) was associated with PIP. Model 3 demonstrated several variables that were statistically associated with PIP, including lower educational levels (OR 1.56, CI 95% 1.03–2.43 p = 0.047) and taking hypoglycemic agents (OR 9.27, CI 95% 5.06–16.98, p < 0.001), nervous system disease drugs, including psycholeptics, psychoanaleptics and antiepileptics (OR 25.66, CI 95% 13.54–48.63, p <0.001); antiasthmatic drugs (OR 5.18, CI 95% 1.35–19.80, p = 0.016); gastrointestinal disease-related drugs (OR 4.92, CI 95% 2.10–11.56, p <0.001); and anti-inflammatories, antirheumatics and analgesics (OR 278.07, CI 95% 152.61–505.66, p <0.001) (Table 5).

Table 2 Chronic diseases and drugs by self-reporting
Full size table
Table 3 Potentially inappropriate prescribing, according to Beers criteria (2015)
Full size table
Table 4 Risk factors for having a potentially inappropriate prescription
Full size table
Table 5 Multivariate logistic regression analyses for having a potentially inappropriate prescription
Full size table

Discussion

The use of inappropriate medications in older patients is a public health problem that affects the morbidity and mortality of this population. The results of this study show that certain groups of drugs have higher probabilities of being inappropriately prescribed to elderly patients, regardless of clinical conditions such as frailty, cognitive impairment or functional dependence, conditions that make patients more vulnerable to suffering negative clinical outcomes (falls, hospitalizations, disabilities, and death) [21]. In addition, the strong association of polypharmacy with PIP documented raises this risk even more.

A relevant finding is that almost half of the patients were exposed to PIP, which is higher than the rate reported by other authors [22, 23]. Therefore, avoiding inappropriate medications should be a primary objective of medical attention for the elderly [3,4,5]; selection, dose and duration of pharmacological treatments should be carefully evaluated in clinical practice, particularly in patients suffering cognitive impairment (including the use of central action drugs that often produce chronic or acute behavioral and cognitive alterations) [24].

Regarding specific medications the results also showed that, according to the Beers Criteria, prescriptions for NSAIDs and glibenclamide are the most frequently identified PIPs. This pattern differs slightly from other reports that have found that hypnotic, sedative and anxiolytic drugs are the most frequently identified, present in more than two-thirds of PIP [22, 23, 25,26,27,28,29,30]. In Mexico, glibenclamide is included in the reference drug list of the IMSS; as a result, it is frequently prescribed to vulnerable elderly patients. Something similar occurs in Brazil, where this drug is included in the list of essential medicines, and the Brazilian government distributes it in primary care facilities [31]. Therefore, improving practitioners’ prescription quality should include actions related to the medications that are sold or included in health institutions’ medication formularies or the reference drug list.

Proton-pump inhibitors are high-risk medications that were found associated to PIP. The belief that these drugs are simply “gastric protection” has increased their use, and their treatment is often prolonged. Appropriate evaluation for bleeding risk factors before prescribing these drugs should be essential, as well as continuing periodic clinical and prescription review to define the need to discontinue their use. The risk of serious adverse events, such as community-acquired pneumonia, bone fractures, Clostridium difficile infection, hypomagnesemia and kidney damage, is well known [32]. This finding is consistent with the global increase in the prescription of these medications, and points to a clear need to create strategies that guide doctors in their correct use, such as technology and multifaceted educative interventions, among others [33,34,35,36].

Older adults have high rates of pain syndromes and osteoarticular disease, for which anti-inflammatory-antirheumatic and analgesic medications (ATC M01 and N02) are frequently prescribed. The elevated proportion of patients with PIP who were prescribed these medicines is noteworthy in this study, and the proportion identified is higher than that mentioned in the literature (e.g., 18% in Brazil [37] and 31.4% in Portugal [22]) The overprescribing of NSAIDs among the patients in Mexico is a problem that has been repeatedly reported in the literature [10, 38]. It is also a problem for which the health system has not yet incorporated strategies to contain, regardless of the existing evidence of effective interventions in this context [39, 40].

Antiasthmatic drugs (R03) were also associated with PIP, with theophylline being one of the most frequently prescribed drugs of this group. This PIP illustrates the risk to which patients are exposed: theophylline affects the central nervous system (insomnia) and raises the risk of arrhythmias. Finally, other examples of PIPs identified in this study include the use of anticholinergic drugs that reduce urine flow and cause urinary retention, the use of nonselective beta-blockers in patients with chronic obstructive pulmonary disease (COPD), and the use of systemic corticosteroids instead of inhaled corticosteroids for the treatment of patients with moderate-to-severe COPD [14, 41].

Among sociodemographic characteristics that showed association with PIP, low educational level has also been described in other studies [26,27,28]. This finding suggests that patients’ levels of awareness of their health and treatment, as well as their expectations, may play relevant roles in the prescriptions they receive. Patients with higher educational levels may have better access to information about medications and thus be more active in doctor-patient communication, exercising an influence over their prescription. Our results confirmed that even when controlling for the number of diseases, the effect of education level with respect to PIP held steady, as another study also showed [26]. This suggests that patients’ educational levels can influence the quality of the medications they receive, independent of their clinical conditions, and highlights the importance of considering educational levels when designing intervention strategies to improve drug prescription to older adults.

A strength of this study is that the findings show the need in Mexico for having an updated reference drug list containing drugs that have been shown to be effective, safe and at reasonable costs; furthermore, the inclusion of the pharmacist in the health teams, and the modernization of the clinical record with electronic alerts for enhance medication use should be considered as part of the national drug policy.

Some limitations must be recognized. The secondary data analysis did not allow the application of more sensitive PIP criteria such as STOPP/START [42]. In addition, although these results are not generalizable to other contexts, they are representative of primary care users in a public institution that covers nearly 50% of the population of Mexico [43].

Conclusion

This study complement information regarding the potentially inappropriate use of medications in ambulatory healthcare provided to elderly patients. The significant association of medical and nonmedical factors with PIP highlights the need to take actions to improve the quality of drug treatment, through strategies directed both at the education of patients and the training of doctors in geriatric pharmacotherapy. Including the participation of the pharmacist in the care of elderly patients, computer support for decision-making in clinical practice should also be considered in those Countries in which these strategies have not been implemented. Additional actions must seek to reduce the risks of adverse health events by strengthening effective, educational communication with elderly patients.