Answers are grounded in this paper and cite their sources.
SIR—Delirium remains a common, serious and under-recognised problem affecting older hospitalised patients. Despite being associated with poor longer-term outcomes, including death and institutionalisation [1], delirium remains disproportionately ignored relative to impact [2], and poor recognition remains the greatest obstacle to improved care and research [3]. It is ‘missed’ in up to two-thirds of cases [1]. Under detection of delirium may occur for a number of reasons; symptoms are heterogeneous and transient [1], and diagnosis is subjective relying on clinical skills in the absence of a diagnostic ‘test’ [2]. There is conflicting data on whether correct diagnosis is associated with demographic factors (male gender and older age [4]) or ethnicity [5]. Poorer English and education may decrease detection [6]. ‘Hypoactive’ delirium [7] with its less dramatic presentation and psychiatric co-morbidity may impede correct diagnosis, possibly through incorrect attribution of a psychiatric diagnosis [8, 9]. There is little published UK data, and previous work is limited by small numbers and under-reporting of co-morbid conditions [1]. We aimed to identify the patient characteristics associated with a correct detection of delirium in older patients with unplanned acute medical admissions. Subjects were recruited from a large north London general hospital (serving a primary care trust (PCT) population of almost 1.5 million) over 6 months (4th June 2007 to 4th December 2007). All patients aged over 70 with unplanned acute admission to the Medical Acute Admissions Unit (MAAU) were eligible. The MAAU takes all medical admissions direct from the accident emergency department and general practitioners (GPs), except those requiring urgent coronary artery procedures, acute stroke patients and those admitted to surgical specialities. Participants were excluded if they were admitted for <48 h or did not speak sufficient English for assessment with the Confusion Assessment Method (CAM) [10]. All patients were assessed within 72 h of admission by an old age psychiatrist or specialist registrar. Subjects were screened with the CAM which has high sensitivity and specificity for the detection of delirium and discriminates between delirium and dementia. It can be completed in <5 min and consists of nine operationalised criteria for delirium from the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R) [10]. All assessors used the same standardised version that maximises sensitivity [11]. The primary outcome was the correct detection and documentation in nursing or medical notes of delirium by the medical team (compared to detection by the study team using the CAM) within the first 72 h of admission. Pseudonyms acceptable for delirium were ‘acute confusional state’, ‘acute confusion’ and ‘delirious’. ‘Confused’ and ‘disorientated’ were not accepted for diagnosis of delirium in the main analysis, although we subsequently conducted a sensitivity analysis, accepting more approximate terms as evidence of detection. We then compared patients with detected and non-detected delirium in relation to a number of clinical and demographic characteristics (see Table 1). These included: demographics, whether the admitting team had attempted screening for cognitive impairment using the Abbreviated Mental Test Score (AMTS) [12], the severity of acute illness (using the Acute Physiology And Chronic Health Evaluation (APACHE) II score [13]) and chronic co-morbidity (calculated using the Charlson Co-Morbidity Index (CCI) [14]), continence, risk of pressure sores (Waterlow score) and the principle cause of admission (obtained from Hospital Episode Statistics using the ICD-10 diagnosis, which were recorded independently by the hospital coding department and categorised according to the Ambulatory Care Sensitive Condition system [15]). We also considered whether routine blood results and whether the individual ‘agitation’ item of the CAM predicted correct diagnosis of delirium. Characteristics of patients with detected and undetected delirium Bold P values above indicate characteristics which were significantly different (P <0.05) in patients with undetected delirium compared to those with detected delirium. Data missing on three subjects. Item from CAM. Characteristics of patients with detected and undetected delirium Bold P values above indicate characteristics which were significantly different (P <0.05) in patients with undetected delirium compared to those with detected delirium. Data missing on three subjects. Item from CAM. We sought verbal consent from participants or, if they lacked capacity, verbal assent from family carers or their key nurse. Screening for cognitive impairment, dementia and delirium should be a routine clinical practise [16]. The exclusion of patients unable to give consent or without a relative to give assent may have caused selection bias, excluding the patient population we wished to study [17]. The study was approved by the Royal Free Hospital NHS Trust Ethics Committee, and the methodology is described in full in the parent study by Sampson et al. [18]. During the study period, 805 eligible patients were admitted, of these, 45 (5.5%) were missed (untraceable or public holidays), 30 (3.7%) refused to participate and 20 (2.5%) had insufficient English. Thus of all admissions, 710 (88.2%) were screened for delirium. The mean age of subjects in the cohort was 83 (range 70–101), and 462 (59%) were female. Using the CAM, 110 (15.5%) of the cohort were delirium cases. Of these, 79 (72%) were not detected by clinical teams. The table compares characteristics of those with and without detected delirium. Patients with undetected delirium were slightly younger (85.5 vs 87.5 years), more likely to be unmarried, live in a nursing or residential home and be of white British ethnicity. Prior psychiatric or dementia diagnoses and assessment with the AMTS by the admitting team were not associated with detection of delirium. Patients with undetected delirium had higher levels of co-morbidity (CCI) and incontinence, and significantly fewer were admitted with urinary tract infection (UTI). There were no significant differences between the two groups in relation to clinical measures (glasgow coma scale (GCS), APACHE scores, white cell count or albumin). Patients with undetected delirium had lower median C reactive protein (CRP) and platelet levels. The sensitivity analysis that accepted more approximate terms revealed an improved detection rate of 40% but otherwise the same pattern of results; 27.7% of those with undetected delirium had a CCI score of ≥5 compared to 2.3% with detected delirium (Χ2 = 8.89, P = 0.002), and 9.2% with undetected delirium had a UTI compared to 23.3% with detected delirium (Χ2 = 4.03, P = 0.045). We examined data from a large, representative cohort of older people. Our observed point prevalence of delirium in general medical admissions (15.5%) is consistent with previously reported figures of 15–20% [19]. This study observed a higher rate of undiagnosed (72%) delirium compared with previously described non-detection rates of between 33 and 66% [1] despite the fact that we only measured single point prevalence. Our rates are closer to those found by Elie et al. [6] who reported a high non-detection rate of 65% within a busy emergency department. This may have occurred because most previous studies were conducted in general medical wards. This study was conducted in a busy acute admissions unit to which patients were rapidly transferred from the emergency department and then onto a ‘base ward’ (depending on diagnosis). We hypothesise that this rapid transfer of patients, often through at least three settings in 3 days, makes the diagnosis of delirium even more of a challenge. However, given that undocumented delirium was assumed to equate undetected delirium in this study, the observed higher rate of undiagnosed delirium may also reflect poorer documentation standards in busy clinical settings. Several statistically significant differences between patients with detected and undetected delirium were observed. Patients with undetected delirium had a greater burden of medical co-morbidity on the CCI. It may be that a more extensive history of medical co-morbidity distracted treating medical teams from diagnosing delirium. However, the similar APACHE scores observed for both these groups suggest that this diagnostic clouding was not related to acute physiological changes. Patients with detected delirium were more likely to present with a urinary tract infection. This could reflect traditional undergraduate teaching on delirium where UTI is a key differential diagnosis. It may also reflect over-diagnosis of UTI in frail, older patients with coincidental asymptomatic bacteriuria [20]. Patients with undetected delirium were more likely to be doubly incontinent prior to admission compared to detected patients, and this approached statistical significance (P = 0.061). This may reflect the general frailty of these patients. We were unable to replicate some findings from previous work. A co-morbid psychiatric history was not more frequent in undetected delirium patients (although it is acknowledged that this may have been under-documented on admission). This was in contrast to previous studies [8,9] where psychiatric co-morbidity impeded correct diagnosis. It is thought that hyperactive forms of delirium are better detected than hypoactive subtypes [7]. However, we found that the presence of marked psychomotor agitation did not improve the diagnosis of delirium. Where the admitting team carried out the AMTS, there was a slightly higher rate of detected delirium (55 vs 35%) but this result was not significant. There was a tendency for subjects with undetected delirium to be older, unmarried and from ethnic minority groups, and this has been found in other studies. Laboratory findings were not found to be significantly different in detected and undetected delirium groups. Detected patients were observed to have a higher median CRP, but this result was not significant. Our cohort was large, but our study may still have been underpowered to detect differences between the two groups. Our case note review relied on clinical notes where undocumented delirium was equated to undetected delirium. The study design could not accommodate situations where delirium may have been detected but not documented. The robustness of admission diagnoses (such as UTI) extracted from hospital episode statistics and based on ICD-10 coding could also challenged. We have demonstrated how delirium remains undetected in a typical general hospital, and detection is strongly influenced by the admission diagnosis. Other demographic characteristics may influence the diagnostic process. Detection and management of delirium in acute hospitals is now thought to be an important quality issue [21]. The routinely used test on acute medical admission, the AMTS [12], does not significantly improve delirium detection rates (and it is not designed to do this). We would suggest that only when specific tools for delirium screening, such as the CAM are incorporated into routine practice, as suggested by clinical guidelines will detection improve. Prevalent delirium remains poorly detected in older medical inpatients. Prior psychiatric or dementia diagnoses and assessment with the AMTS by the admitting team were not associated with detection of delirium. Patients with undetected delirium were more likely to display high levels of medical co-morbidity, incontinence and significantly less likely to have an admission diagnosis of UTI. E.L.S. was funded by a Medical Research Council (UK) Special Training Fellowship in Health Services Research. We thank Dr Dan Lee (Health Services for Elderly People, Royal Free Hospital NHS Trust), the staff of the Medical Acute Admissions Unit and Ginnette Kitchen, Lucy Watkin, Jenny Drife and Viv Green for assistance with data collection and processing. There is no conflict of interest to declare.
No takes yet. Share an insight, caveat, or question.
Collins et al. (2009) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: