Prevalence and risk factors of delirium diagnoses in primary care practices

The term delirium describes an etiologically unspecific organic brain syndrome that is characterized by simultaneous disorders of consciousness and attention, perception, thinking, memory, psychomotor skills, emotionality and the sleep-wake rhythm. The duration and severity can vary greatly [Dilling H et al., 2014]. Delirious syndromes are a common complication in the inpatient treatment of elderly patients. While delirious syndromes with a prevalence of 1-2 % were rarely found in outpatients over 65 years of age (Lange et al., 2013; Davies et al., 2013), there were sometimes very high prevalence rates of up to 70% in nursing homes (between 1.4% and 70.3%) and up to 82% in hospitals (between 7% and 82%) (Inouye et al., 2014, Lange et al., 2013). The wide range of these findings depends primarily on the diagnostic criteria used, the age of the patient and the dementia rate. In general, delirium is associated with increased length of stay, increased care needs, increased institutionalization, and ultimately increased mortality rates (Inouye et al., 2014; Pendleburry et al., 2015). It is known that delirium often goes undetected in both inpatient and outpatient settings (Lange de, et al. 2013, Jenkin et al., 2016). Therefore, identifying risk factors is particularly important in order to initiate preventive measures as early as possible (e.g., Kratz et al., 2015; Pendlebury et al., 2017). This applies in particular to preventive diagnostic instruments, such as those used, for example. B. were introduced for delirium in dementia (Richardson et al., 2017). However, there is little information on the prevalence and risk factors of delirium in the outpatient care sector (Lixouriotis et al., 2011; Lange et al., 2013). The aim of this study was therefore to determine the prevalence of documented delirium diagnoses in a large sample of patients in German general practices and to investigate whether risk factors that make the occurrence of delirium more likely can be identified.

>> This study is based on data from the Disease Analyzer database (IQVIA). This collects information on drug prescriptions, diagnoses and basic medical and demographic data, which is provided directly and in anonymized form by the computer systems in doctor's offices (Dombrowski and Kostev, 2017).

study population
The study sample included patients aged 65 and over who had received a delirium diagnosis (ICD 10: F05) for the first time from a German family doctor between January 2012 and December 2016 (index date). Of 1,261 general practitioners, 851 documented such a diagnosis for at least one patient. Patients with an observation period of less than 365 days before the index date were excluded. After applying identical inclusion criteria, delirium patients and control patients without delirium were matched based on age, gender and doctor (1:1) (Fig. 1).

variables
Age, gender and type of accommodation (at home or in a nursing home) were collected as demographic data. Clinical data included the following diagnoses documented within 365 days before the index date: hypertension, obesity, hyperlipidemia, coronary artery disease, heart failure, myocardial infarction, stroke including TIA, diabetes, renal insufficiency, liver fibrosis and cirrhosis, cancer, superficial injuries, open wounds, Fractures, depression, anxiety disorders, adjustment disorders, dementia of any origin, epilepsy, Parkinson's disease, sleep disorders, hearing loss, blindness and visual impairment, senility, urinary tract infections, upper and lower respiratory tract infections, urinary incontinence and fecal incontinence.
In addition, the following drug classes documented within 365 days before the index date were available for analysis: non-lithium antidepressants, neuroleptics, benzodiazepines, lithium, anti-parkinsonians, antiepileptics, analgesics, corticosteroids, diuretics, beta-blockers, calcium channel blockers, antibiotics, Antiarrhythmics, antihistamines, anticholinergics, antiasthmatics and muscle relaxants (M03). In addition, the total number of medications prescribed within one year before the index date was analyzed.

Statistical analyses
Descriptive analyzes were performed for demographic and clinical variables. The mean ± SD was calculated for age, while the respective proportions were determined for the other variables. A logistic regression model was used to calculate the association between delirium and the different variables. Bonferroni correction was applied to counteract alpha error accumulation in multiple comparisons. For the 47 variables in the regression model, p values <0.001 (calculated as 0.05/47) were considered statistically significant. Analyzes were performed using SAS version 9.3.
Results
Among the 6,180,042 patients who had visited one of the 1,262 general practitioners in Germany at least once between January 2012 and December 2016, there were 4,792 patients in 851 practices for whom a diagnosis of delirium had been documented. This results in a five-year prevalence of 0.08 % and an average of 0.7 (SD: 1.5) patients per practice per year. Of the 4,792 patients, 4,444 had a first-time diagnosis of delirium (no diagnosis of delirium before the index date).
Patient characteristics are shown in Table 1. The study included a total of 2,194 patients with a diagnosis of delirium and an observation period of at least 365 days before the index date and 2,194 control patients. The average age was 82.2 years (SD=7.1 years) and 43.0 % participants were men. Approximately 24 % of the delirium patients and 10 % of the control patients lived in nursing homes.
Delirium was positive with nursing home placement (OR=1.69), dementia (OR=3.45), epilepsy (OR=2.16), stroke (OR=1.80), Parkinson's disease (OR=1, 78), sleep disorders (OR=1.64), renal insufficiency (OR=1.57) and fractures (OR=1.56). Delirium was also associated with the prescription of benzodiazepines (OR=1.62) and antiepileptic drugs (OR=1.53). Finally, there was also a positive association between the risk of delirium and the number of different drug classes prescribed in the year before the index date. Compared with patients without drug treatment, the OR for delirium was 1.53 for one drug class, 1.83 for two classes, 1.97 for three classes, 2.22 for four classes, and 3.21 for more than four classes.
discussion
Prevalence
The five-year prevalence of documented delirium diagnoses in German general practitioners' practices was 0.08 % in the study presented here. It is particularly important that in our sample only 851 of the 1,262 general practitioners (67.4%) even documented a diagnosis of delirium. It can be assumed that a third of the general practitioners were not very familiar with the diagnosis of delirium. Delirious syndromes are probably underdiagnosed in both inpatient and outpatient care settings. The knowledge and attitude of physicians regarding the diagnosis of delirium should therefore be improved (Jenkin et al., 2016). This is particularly true for the diagnosis of hypoactive delirium, as the diagnosis is particularly difficult due to the lack of immediately obvious symptoms (Peritogiannis et al., 2015). It can be assumed that delirium diagnoses in German general practitioners' practices were usually made only on the basis of clinical symptoms without using standardized assessment methods (Grover and Kate, 2012; van Velthuijsen et al., 2016). This could reflect the difference between outpatient and inpatient care, as standardized examination procedures are increasingly being used in inpatient care as part of delirium prevention (Kratz et al., 2015; Pendleburry et al., 2017).
Another reason may be that doctors do not use an ICD-10 code intended for delirium when diagnosing delirium. It is conceivable that delirium in dementia is coded with a dementia code (e.g. G30) and not with a delirium code (F05). In addition, it is conceivable that many doctors, especially those who are not trained in ICD 10, diagnose delirium (similar to alcohol delirium) as productive symptoms in the context of psychosis and not as temporary confusion in the context of delirium. This could also be a reason for the underestimation of the frequency of non-substance-related delirium in general practitioners' practices.

Home accommodation
Patients living in institutions had a higher risk of developing delirium. In the German healthcare system, general practitioners treat both patients who live at home and patients who are housed in a facility. In general, institutionalization is associated with significant morbidity-related impairment of everyday abilities (Arinzon et al., 2011).

dementia
In several reviews, dementia is considered a major risk factor for the development of delirium (Inouye et al., 2014; Oh et al., 2015; Yang et al., 2017). The results are comparable to those found in our study. The risk analyzes of population-based studies also identified the occurrence of cognitive disorders and dementia as an important risk factor for the development of delirium (Lange et al., 2013).

Other neuropsychiatric disorders
In addition to dementia, other neurological and psychiatric diseases such as epilepsy, stroke, Parkinson's disease and sleep disorders also have an increased risk of delirium in patients treated by primary care physicians. Individual studies have shown that epilepsy (Neije et al., 2014; Woodford et al., 2015), sleep disorders (Terzaghi et al., 2014), Parkinson's disease (Vardy et al., 2015) and stroke (Shi et al ., 2012) are associated with an increased risk of delirium. However, with the exception of stroke, neither the reviews of population-based studies (Lange et al., 2013) nor the reviews of inpatients (Achmed et al., 2014; Inouye et al., 2014) showed evidence of an increased risk of delirium in connection with the aforementioned diseases.
The prescribed benzodiazepines and antiepileptic drugs represented further risk factors. The prescribed drugs are associated with the neuropsychiatric diagnoses of sleep disorders and epilepsy.

Internal and surgical diagnoses
In addition to neuropsychiatric risk factors, only renal insufficiency and fractures were identified as risk factors in the areas of internal medicine and surgery. Renal insufficiency is a clinically recognized risk factor for delirium. To the best of our knowledge, no comprehensive risk studies have been carried out on this subject. Dialysis patients in particular often suffer from psychiatric comorbidities. These comorbidities include depression, anxiety disorders, suicidality and delirious syndromes (De Sousa, 2008). It is possible that the patients with renal insufficiency who were treated by family doctors were also treated by specialists who then e.g. B. a diagnosis of delirium was made during dialysis treatment without the family doctor also coding this diagnosis. Regarding surgical diseases, our results are consistent with the information regarding hip fractures, which commonly occur in advanced age. Recent reviews of hip fractures have demonstrated the importance of such fractures in the development of delirium (Oh et al., 2012; Young et al., 2017). The patients in our study were probably those who had been treated by a family doctor as part of inpatient fracture treatment postoperatively after delirium, as well as those who had been discharged from the hospital prematurely with protracted delirium (cf. Peritogiannis et al., 2015; Pendleburry et al., 2015).

Polymedication
Finally, the number of different drug classes prescribed within 1 year before the index date was positively associated with the risk of delirium. Comparable results have been reported for the inpatient setting (Ahmed et al., 2014; Inouye et al., 2014). Of particular interest in our sample was the fact that only a higher number of prescribed substances was associated with a higher risk of delirium, but not individual, potentially delirium-inducing substances selected for this study, which are also referred to as “high-risk drugs” (Ahmed et al. 2014). This result is likely due to the sample size and differences between inpatients and outpatients. The increased risk when administering more than one potentially delirogenic drug is probably due to a summation effect, increased pharmacokinetic or pharmacodynamic interaction potential, or administration errors resulting in overdose that can occur in the context of multimedication in multimorbid and cognitively impaired patients (Alegiakrishnan and Wiens, 2004 ; Inouye et al., 2014).

Limitations and strengths
This study is based on an evaluation of routine data that is generated as part of the documentation requirement and for billing with health insurance companies. In the German healthcare system, the ICD-10 (Dilling et al., 2014) is used as the basis for relevant documentation. The diagnostic quality cannot be compared with the data from scientific studies. Information about how primary care physicians diagnosed delirium was not available. It can be assumed that primary care physicians made delirium diagnoses based on clinical symptoms without performing further tests. The diagnosis may also have been taken from the documents prepared by the hospital (discharge report) or by specialists (consultation reports, doctor's letters). It is very likely that the rate of delirium-specific tests in Germany is as low as tests in dementia diagnostics (Eichler et al., 2015). In addition, specific delirium tests (Grover and Kate, 2012) are not included in the fee schedule, which regulates the remuneration of medical services in the outpatient sector.
One of the strengths of the study is the size of the original sample, which included more than six million patients and more than 1,200 primary care practices. The largest sample of patients (living at home or in institutions) studied for the incidence of delirium included N=10,262 participants (Andrew et al., 2006). The analyzes of insurance data included just over 75,000 patients. However, these data took into account all health areas (Fick et al., 2005).
conclusion
The diagnosis coding of delirium is a very rare event in primary care. Given the low prevalence and the high proportion of practices that do not use this diagnosis, it can be assumed that a large proportion of physicians were unfamiliar with the diagnosis of delirium. The knowledge and behavioral routines involved in making a diagnosis therefore need to be improved in quality.
Compared to inpatient treatment, neuropsychiatric risk factors, especially dementia and the number of high-risk medications prescribed, were particularly important for the diagnosis of delirium in German general practices. The reasons for this need to be further investigated. The question of whether listing these specific diagnoses and medications would be useful in the context of interface management of hospital admissions by primary care physicians to inform the receiving hospital of the risk of delirium also requires further investigation. The methodological limitations of analyzing data from routine care must be taken into account. <<