Hospital coding of Dementia: is it accurate?

Hospital coding of Dementia: is it accurate? Elizabeth Cummings, Roxanne Maher, Christopher Morris Showell, Toby Croft, Jane Tolman, James Vickers, Christine Stirling, Andrew Robinson and Paul Turner Abstract This paper investigates the coding of dementia in the episode of care in a pilot study group (N=48) post hospital discharge and the possible implications of under-coding. The assigned ICD-10-AM codes and Diagnosis Related Groups were reviewed. Results demonstrate under-coding of dementia and of cognitive defi cits; poor correlation between admission diagnoses and dementia codes on separation; and changes in individual patients cognitive status across forms and assessments in the same admission. The complexities of accurately coding dementias will impact upon planning for future treatments and service provision and will have a fl ow-on effect for patients, hospitals, and patient care in Australia. Keywords (MeSH): Dementia; Clinical Coding; Database Management Systems; International Classification of Diseases; Health Services Research; Informatics; Medical Records Supplementary keyword: Health Information Management Introduction Australia currently has more than 250,000 cases of dementia, with a predicted increase of 327% in the next 40 years (Alzheimer s Australia 2008). People diagnosed with dementia can be expected to live on average four to eight years post diagnosis, depending on their age and the stage of the dementia at diagnosis (Alzheimer s Australia 2008). Figures also suggest that dementia rates as the fourth leading cause of death in those over the age of 65 years, and it is anticipated that by 2016 it will be Australia s largest contributor to the disability burden. A defining feature of dementia is cognitive impairment (Folstein 2007). As a result, hospitalised dementia patients may be less able to raise alarms about their condition because of the lack of cognition regarding their health status (Fillit et al. 2002). Psychiatric issues, including depression, delusions, hallucinations, aggression and irritability, are also frequently associated with dementia (Wancata et al. 2003), with some reports suggesting that more than 50% of delirium also occurs in patients with dementia (Fillit et al. 2002). In light of the poor prognosis of the condition and the potential benefits of treatment, early detection of dementia is desirable (Chang et al. 2009; Laurila et al. 2004). However, diagnosis of dementia is usually initiated in a primary care setting, based on patient symptoms or caregivers concerns (Brayne, Fox & Boustani 2007). Due to the brevity of doctor-patient interactions and patients presenting with multiple problems, diagnosing dementia can be challenging. Unless directly assessed, the early, subtle symptoms of dementia, such as memory impairment, may not be apparent (Boise et al. 1999). Therefore, many studies have found delayed or undocumented diagnosis of dementia (Callahan, Hendrie & Tierney 1995; Iliffe et al. 1991; McCormick et al. 1994), which can compromise patient safety, treatment compliance, and recognition of co-morbidities (Chodosh et al. 2004), as well as increased patient and caregiver burden (Bradford et al. 2009). Patients require care that is tailored to their specific cognitive deficits and reflects the expected health outcomes (Park et al. 2004). This can happen only if dementia is detected and documented at admission or early in the hospital stay. Healthcare workers face significant challenges in the management of patients with dementia, which will increase as the prevalence of the disease increases (Brookmeyer, Gray & Kawas 1998). In financial terms, it has been found that the medical management of people with dementia results in higher healthcare costs (Kinosian et al. 2000; Taylor & Sloan 2000; Gutterman et al. 1999; Albert et al. 1999). This may be directly related to the dementia or to coexisting comorbidities due to poor compliance with medications, diet, and other self-management requirements. Studies also reveal that patients with dementia tend to have more frequent and longer hospital stays and an increased number of complications, which in turn require greater use of resources (Fillit et al. 2002; Hill et al. 2002). It is undeniable that the increased incidence and impact of dementia places a great burden on society in general, and health resources in particular (Kukull et al. 2002; Chodosh et al. 2004). There is also a range of common medical conditions affecting cognition in older populations that needs to be distinguished from dementia by clinicians and coders. These include stroke, delirium, hyponatraemia, vitamin deficiencies and other malnutrition issues, concussion/ post-traumatic amnesia, poly-pharmacy issues affecting cognition, alcohol withdrawal, normal pressure hydrocephalus, hypothyroidism and thiamine deficiency (Attix & Welsh-Bohmer 2006). It is often difficult for hospital medical staff to make a differential diagnosis in a confused elderly person immediately upon admission, as many of these acute (and potentially reversible) condi- HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE) 5

tions must be excluded before a diagnosis of dementia can be made. Medical records of such patients will frequently contain multiple cognitive assessments, blood results, urine analyses, CT scans and similar over many days as medical teams narrow down the range of possible conditions accounting for cognitive impairment and functional failure leading to admission. From the perspectives of the healthcare professional, Health Information Manager (HIM), researcher and patient, it is important that accurate, reliable and relevant information is captured, stored and researched. This will enable future health service planning, resource allocation, formation of policies and trialling of treatments (Zilkens et al. 2009). International studies have reported misclassification and under-coding of dementia and related conditions (Greco et al. 2005) in acute wards, due to the limitations of both the International Classification of Diseases and Related Health Problems, Ninth Revision, Clinical Modification (ICD-9-CM) (Fillit et al. 2002) and failure to diagnose. This can also be true of the International Statistical Classification Of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM) when diagnoses that do not fulfil the requirements of Australian Coding Standards (ACS)-0002 Additional Diagnoses are not coded. All state and territory health authorities must collect morbidity statistics that come from coded patient record data. Australian Coding Standards (ACS) apply to every hospital in Australia and provide the clinical coder with guidelines and standards to aid in assigning ICD-10-AM codes to all hospital separations. The two main standards that impact on whether a dementia code is recorded or not when a dementia patient is admitted to hospital or residential care facility or has an attendance at a healthcare facility are: ACS 0001 (Principal Diagnosis [PD]), and ACS 0002 (Additional Diagnosis [AD]). Clinical coders are not responsible for identifying the PD. This is the role of the medical practitioner caring for the patient and is usually selected at separation and written on the discharge summary or front page of the patient s admission notes. HIMAA (2006) suggested that clinicians are not always accurate in selecting the correct PD because they are not always fully acquainted with the definitions. For this reason, clinical coders must be aware of their professional obligations to choose an appropriate PD when coding and consult the clinician when necessary. The definition for a PD cited by the National Centre for Classification in Health (NCCH) (2008:10) is that provided by the Health Data Standards Committee s National Health Data Dictionary, Version 13, (AIHW 2006): the diagnosis established after study to be chiefly responsible for occasioning an episode of admitted patient care, an episode of residential care or an attendance at the health care establishment as represented by a code. Clinical coders must do a thorough search of the medical record to establish a suitable PD code. In applying this rule, coders may evaluate the results of MMSE, PAS, and Cornell scales, in addition to allied health professional and doctors case notes, tests and assessments. However, a clinical coder may not code a diagnosis where a diagnosis has been made. It is the role of the clinicians to document their findings and conclusions following the completion of such assessments. In comparison, assigning an AD in line with ACS 0002 guidelines is much more complicated and problematic for clinical coders. An AD, according to the Health Standards Committee and cited in NCCH (2008:13), is defined as: [a] condition or complaint either co-existing with the principal diagnosis or arising during the episode of admitted patient care, episode of residential care or attendance at a health care establishment, as represented by a code. The intention of this standard is to determine the type and extent of care provided to inpatients in Australian hospitals for the Admitted Patient Care National Minimum Data Set. To this end, NCCH advised coders that AD allocation should be made when a patient s condition requires any of the following: commencement, alteration or adjustment of therapeutic treatment diagnosis procedures increased clinical care or monitoring (NCCH 2008:13). By this standard, any patient who is diagnosed with dementia or cognitive impairment, or who has treatment altered for such a condition during the episode of care, ought to have an AD of the specific condition. The standard s final criterion (above) also provides for the allocation of an AD code when these conditions are present in a patient without any changes to treatment being necessary. For clinical staff experienced in caring for patients with any form of dementia or cognitive impairment, it is obvious that increased clinical care or monitoring is necessary. Unfortunately, in some circumstances the clinician may not document in such a way as to enable the coder to then allocate codes and, in addition, clinical coders may not have the same understanding of the care requirements of the conditions. In some circumstances, the clinical coder must consult with the medical officer who completed the discharge documentation to clarify the case and possibly over-ride the PD made by that person. The standards pertaining to mental and behavioural disorders (F codes), under which dementia and cognitive impairment are classified, contain only two brief standards specifically applying to these conditions: ACS 0528 (Alzheimer s Disease) and ACS 0532 (Cognitive Impairment). The former states that, for Alzheimer s disease, dementia can always be assumed and coded. The latter standard makes minimal contribution in assisting the coder to allocate a code in the event of poor documentation. 6 HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE)

Method Participants A retrospective, observational prevalence study was undertaken on a pilot group of patients admitted to two hospital wards, predominantly by transfer from general wards, during the trial period 1 January 2009 to 31 March 2009. Ethical approval for this research was gained through the Tasmanian Human Research Ethics Committee (Project H10215). All patients in the pilot, or a legal person responsible consented for their data to be collected and their records examined. Data were not collected on non-consenting patients. During this period, there were 49 consenting admissions, representing 45 individual patients: three patients had two admissions during this 3-month period; however, in terms of evaluating documentation these were treated as unique admissions. One patient (SN 500) remained an inpatient at the end of the study period and therefore had not been allocated any separation codes; data from that patient have been excluded from the analysis. Thus, the complete data set comprised 48 admissions, referred to as the study cohort. There were no specific inclusion/exclusion criteria for the study and demographic data were not recorded as part of this analysis. The cohort consisted of 27 female and 14 male patients. The age range for these patients was 54 to 91 years with a median age of 76 and a mode of 84. Procedure Data from all completed forms were entered into the project database, the Tasmanian Older Persons Database (TOPDB). The database included a number of fields where diagnoses could be documented during an episode of care, but generally these fields related to: the admission diagnosis (diagnosis given at admission) the PD ( the diagnosis established after study to be chiefly responsible for occasioning an episode of admitted care ) ADs (other relevant conditions that arise during episode of care or co-exist with the principal diagnosis and affect patient care) (NCCH 2008). As part of standard hospital practice, salient clinical features of each episode were coded using the ICD-10- AM following patient discharge. Each episode was then assigned to a Diagnosis Related Group (DRG) according to Australian Refined DRGs version 4.2 (AR-DRG 4.2). ICD and DRG data were retrieved from the hospital s casemix system. This study took particular note of coderassigned ICD-10-AM codes from Mental and Behavioural Disorders: F00-F09 (Organic, including symptomatic, mental disorders), referred to here as F-codes. A spreadsheet was used to collate the admission diagnosis; PD; any additional F-code diagnoses; previous admission F-code diagnoses (during admissions over the previous five years); length of stay (LOS); and completion of various inpatient forms and assessments. Data were extracted from the TOPDB and analysed within the spreadsheet to determine the prevalence of dementia as an admission diagnosis or as PD or ADs. The specific DRG code for an admission for dementia is B63Z. Other F-codes that appeared in the database for these patients were also captured. As noted above, there is frequently a close association between dementia and other mental health illnesses such as depression and delirium (Wancata et al. 2003), hence the examination of F-codes. A number of methods are commonly used to assess a person s cognitive capacity. The three main assessment tools used to collect the data included in this study were: Psycho-geriatric Assessment Scale (PAS) Cornell Scale for Depression in Dementia (CSDD) Mini-mental status examination (MMSE) PAS is used to assist health practitioners collect and interpret information on dementia and depression (Jorm & Mackinnon 1997). The CSDD involves interviews with both patient and carer in an attempt to assess the severity of depression in people with dementia (RACGP 2006). Finally, the MMSE is a reliable tool to assess memory, language and visual spatial and executive function (Folstein 2007) in order to identify cognitive impairment. In this test, scores between 24 and 30 suggest normal cognition; scores below 24 suggest subnormal cognition (Corey-Bloom et al. 1995; O'Bryant et al. 2008). The MMSE can also measure deterioration of mental status in already diagnosed dementia patients, who are expected to have a 2- to 4-point decline in scores annually (Folstein 2007). Data from the spreadsheet were reviewed to determine how each set of forms and assessment tools had been used to identify dementia or cognitive impairment within the study cohort. These data were then compared with casemix data and discharge coding to identify cognition assessment tools used for patients discharged with other F-code diagnoses that may be associated with undiagnosed dementia. These scores, of course, are only an indication and the diagnosis of dementia rests upon a comprehensive collaborative history from both patients and carers. Finally, a retrospective review of the individual patient records was undertaken by an informatician and a clinical psychologist who had been involved in the care of these patients. This review was undertaken to understand what issues relating to testing and diagnosis may not have been evident from the data collected in the TOPDB. This review provided the additional evidence described in the results section below. Results Of the 48 aged care and rehabilitation cohort admissions, only 1 patient had dementia listed as their PD, 6 had some mention of dementia in their admission diagnosis, and 5 had a previously un-recognised dementia (not reflected in admission diagnoses) identified and recorded during their admission. However, 2 patients with previously diagnosed dementias did not have these conditions recorded formally anywhere in the records TOPDB HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE) 7

examined for the admission in question. In total, 9 of the 48 patients had a dementia diagnosis recorded in their current admission somewhere (PD, admission diagnosis, and ADs) giving a dementia prevalence for the population of 18.75%. Table 1 summarises these findings. Table 1: Dementia diagnoses recorded by TOPDB on admission and after admission # ADMISSIONS (48 TOTAL) Dementia given as a principal diagnosis 1 Dementia referred to in the admission diagnoses 6 Dementia diagnosed explicitly in admission diagnoses 4 Dementia diagnosed during the admission, not on admission 5 Dementia diagnosed previously, but not currently recorded 2 Examination of these figures clearly indicates that the diagnoses given upon admission is not a reliable indicator of the presence or not of a dementia, as over 5 of the 9 (56%) patients who received a dementia diagnosis received it during the course of the admission rather than on admission. For instance, 1 patient (SN 116) was given the diagnosis of unspecified dementia (F03) after separation, whereas their admission diagnoses included simply Falls/acopia. A second patient (SN 780) who similarly presented with acopia on admission diagnosis was subsequently given a PD on separation of Parkinson s disease (G20), with an AD of dementia in Parkinson s disease (F023). Three additional patients were given dementia diagnoses late in the admission or on separation and had no reference to cognitive impairment on their admission diagnoses. However, Table 1 also indicates that diagnoses of dementia were not always made where it appears they should have been. There were two instances in which patients with previously recognised dementias did not have their dementia diagnosis recorded in any of the documents TOPDB used. For instance, patient SN 537 had a care summary that stated that this patient suffered from dementia, yet no scores for PAS or Cornell were recorded and there was no mention of dementia (or other F-codes) in the admission diagnosis or PD and ADs. This patient was admitted in March 2009 from another ward where F10.1 was coded as an admission diagnosis with specific reference to dementia. In a retrospective search of previous admissions it was found that F01.8 (other vascular dementia) was coded as a diagnosis in January 2007 and again in May 2007. In a second example, patient SN 370 had an admission diagnosis of bilateral pulmonary embolism, chronic pain and Parkinson s disease, but no mention of dementia in the PD or ADs. The rehabilitation assessment noted the patient as being oriented, yet the social work assessment identified that the patient had been diagnosed with dementia after a psychiatric review. The social work assessment also documented poor memory, poor insight, visual hallucinations, paranoia, aggression, confusion, unsettled behaviour including wandering and entering other patients rooms. No MMSE was recorded as having been performed during this stage of care. A retrospective review of previous admissions identified dementia codes (F 023 Dementia in Parkinson s disease) during two previous admissions, including one in late 2008. In fact, this patient was well known to the treating geriatrician and no MMSE was given in the admission because it had been done just weeks earlier in the outpatient clinic. The patient was well known to have a dementia related to Parkinson s disease but this was not in fact recorded formally in the admission records that TOPDB identified. The TOPDB data also indicated that some patients had impaired cognition well recorded formally via MMSE (and similar) in the admission, but with no dementia diagnoses. On first appearances this might indicate under-recognition of dementia by clinicians and/or TOPDB. However, on checking with the medical record it was evident that this is not necessarily the case; many patients had acute medical and or transient impairments to cognition that were not consistent with a diagnosis of dementia. For instance, two patients (563 and 402) had suffered strokes that reduced capacity to perform on the MMSE but which were resolving across time and not indicative of a dementia. A third patient (142) was extremely unwell with a tracheostomy and multiple organ failure; she was essentially palliative and her poor performance on MMSE was not because of a dementia per se. Therefore, it would be wrong for coders to assume that poor MMSEs and similar are necessarily indicative of dementias. Finally, it was clear that many cognitive assessments relevant to dementia were not easily identified by TOPDB and might not be easily identified by coders. For instance, SN 402 was given the Addenbrooke s cognitive exam (which contains the MMSE) with both MMSE and full Addenbrooke s scores reported in the patient medical notes. However, TOPDB did not identify the 15/30 MMSE given here because, perhaps, it was embedded in a form other than the usual stand-alone MMSE. The same patient had detailed speech therapy assessments of cognition (the MAST), which were on the patient records and indicating resolving aphasia, yet these were also unidentified by TOPDB. In another even more concerning example, patient (SN 396) had an admission diagnosis of mild confusion for investigation with no dementia codes allocated following separation. This patient was in hospital for 26 days and during the admission there was no record identified by TOPDB of any cognitive investigations, forms or assessments being completed. In fact, there had been many performed and recorded including an Addenbrookes cognitive exam (containing MMSE) and consultations by a neuropsychologist with Dementia Rating Scales reported in detail. The psychologist concluded there was a possible concussive injury with or without mild cognitive impairment but no dementia. All this input was recorded in the discharge summary, yet 8 HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE)

TOPDB did not identify that any cognitive assessments had taken place. Overall, the documentation related to the forms and assessments was found to be sporadic, with none of the forms used for all patients in the study cohort. The most extensively used forms or assessments were the medical discharge summary and the admission form, which were each used for 25 patients (52% compliance). The nursing admission assessment flagged only one patient as having any cognitive impairment; this was a patient (SN 318) with a documented MMSE score of 21/30, cognitive impairment and short-term memory loss. The social work assessment form appeared to most frequently contain data that highlighted MMSE scores, dementia diagnoses and confusion not documented elsewhere. However, this form was also the most difficult to evaluate as it was primarily free text. Overall, the majority of forms and assessments contained incomplete data sets and so did not appear to present a comprehensive picture of the patient s condition. Depression and delirium are often associated with dementia and may lead to greater risk of permanent nursing home placement and longer length of hospital stay (Wancata et al. 2003). Both may be reversible and for these reasons, conditions such as dementia, depression and delirium can be used as flags to commence assessment of at risk patients for underlying or undetected associated conditions. In the study cohort, 2 of the 9 patients with dementia (SN 327 & 326) were also allocated a depression code on separation (F3290). Another 3 patients had depression mentioned in one or more of the forms and assessments. One of these patients had an MMSE of 23 documented, but no mention of either depression or dementia as a diagnosis. The other 2 patients had no assessment for dementia documented. Two patients were noted as having episodes of delirium during the admission, but no documented assessments for dementia. Discussion This research has assessed the capacity of the TOPDB to capture dementia as an admission diagnosis and used a comparison with ICD coding and DRGs to identify the extent to which data from assessment tools recorded in the TOPDB are able to flag dementia and associated conditions. It has been found that the TOPDB is far from adequate in its present form; it does not highlight all episodes of care where the patient has a diagnosis of dementia, or has been assessed for dementia and found to have some degree of cognitive impairment. The rate of admission diagnosis capture was below 50% for patients having the condition; further work will be required to ensure that these episodes are reliably identified. It was also not possible to reliably identify patients with a dementia diagnosis from a previous episode using the TOPDB. Highlighting patients with dementia or cognitive impairment is both a practical clinical problem and a research problem. This research has also highlighted the issue of patients who are assessed with low MMSE scores at one point in time but no mention of cognitive impairment elsewhere in the record. In two cases, dementia was a co-morbidity that was not investigated further where it should have been, but in other cases (at least three) low MMSE scores reflected a transient problem (e.g. stroke) with no dementia, but this was only evident once the patient medical files were examined in detail. Thus, it is evident that extensive examination of the medical records is necessary to clarify what the TOPDB was unable to capture and it is important to ensure comprehensive medical records to assist clinicians and coders with ensuring the accurate coding. It must be noted here that there are limitations in using ICD-10-AM codes to flag dementia, particularly if the dementia is a co-morbidity. According to the Australian Coding Standards (NCCH 2008), dementia must either be the reason for the patient s admission, or affect the care given during the episode for it to be coded as a PD or AD. In the absence of changes to regular treatment or diagnostic procedures being performed for a dementia patient, this standard means that the Clinical Coder makes the final decision about whether the dementia patient requires increased clinical care and/or monitoring, based on what is documented within the patient record. Prevalence of this condition in a hospital population can be difficult to ascertain with certainty. This aspect of diagnosis coding can be difficult if documentation is poor. Particularly in the case of a patient with mild cognitive impairment, dementia codes may not be able to be used according to the ACS 0002 because there is no specific change to therapeutic treatment and the clinical documentation does not adequately reflect the increased clinical care provided. In this situation, healthcare professionals, HIMs and researchers should not rely solely on diagnosis codes to flag dementia. It is essential to have access to more detailed information that reliably identifies a diagnosis of dementia. Reid, Allan and McIntosh (2005) stress the importance of the accurate documentation and coding of additional diagnoses. Some degree of error with documentation and coding may be unavoidable in the business of healthcare. However, it is essential that the long-term implications of such errors, for all aspects of healthcare, are evident to all those involved in the care process. Conclusion This study of the pilot data collection tool, the Tasmanian Older Persons Database, has revealed an overwhelming discrepancy between data that are collected about dementia and the incidence of dementia and cognitive impairment within this cohort of older hospital patients. Results demonstrate the potential for significant underrecording of dementia and of cognitive deficits, poor HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE) 9

correlation between admission diagnoses and dementia codes on separation, and disparity in individual patients cognitive status between forms and assessments in the same data collection. One factor that has been found to impact upon underdiagnosis of dementia is the classification system that Tasmanian hospitals work within to establish hospital utilisation and influence funding: the ICD-10-AM. Because of the standards used and restricted decisions made by Clinical Coders in coding hospital admissions, dementia is often missed or left undocumented as a coded diagnosis. This results in inadequate information for clinicians, researchers and HIMs to determine the extent of dementia as a diagnosis in these units. Underdiagnosis is also reflected in the DRG allocation. As this lack of completeness impacts planning for future treatments and service provision, it must also have a flow-on effect for patients and patient care. Because of the limitations of ICD-10-AM, a greater reliance must be placed on other clinical data collection tools, and on clinical interpretation of patient history, for the accurate and timely capture of dementia diagnoses in hospital. Unfortunately, the TOPDB in its pilot form failed to deliver the required accuracy and completeness of dementia data within the pilot wards. References AIHW (2006). National Health Data Dictionary. Canberra, ACT, Australian Institute of Health and Welfare. Albert, S., Costa, R., Merchant, C., Small, S., Jenders, R. and Stern, Y. (1999). Hospitalization and Alzheimer s disease: results from a community-based study. Journals of Gerontology Series A: Biological Sciences and Medical Sciences 54: M267-71. Alzheimer s Australia (2008). Dementia: facts and statistics. Available at: www.abc.net.au/rn/.../bbg_20110403_factsfigures australiadementia.pdf (accessed 20 August 2009). Attix, D. and Welsh-Bohmer, K. (2006). Geriatric Neuropsychology. New York, The Guilford Press. Boise, L., Camicioli, R., Morgan, D., Rose, J. and Congleton, L. (1999). Diagnosing dementia: perspectives of primary care physicians. Gerontologist 39: 457-64. Bradford, A., Kunik, M. E., Schulz, P., Williams, S. P. and Singh, H. (2009). Missed and delayed diagnosis of dementia in primary care: prevalence and contributing factors. Alzheimer Disease and Associated Disorders 23: 306-314. Brayne, C., Fox, C. and Boustani, M. (2007). Dementia screening in primary care: is it time? Journal of the American Medical Association 298: 2409 11. Brookmeyer, R., Gray, S. and Kawas, C. (1998). Projections of Alzheimer s disease in the United States and the public health impact of delaying disease onset. American Journal of Public Health 88: 1337-42. Callahan, C., Hendrie, H. and Tierney, W. (1995). Documentation and evaluation of cognitive impairment in elderly primary care patients. Annals of Internal Medicine122: 422-9. Chang, E., Jacoby, S., Wang, J., Aghalar, J., Hussain, R., Pekmezaris, R. and Wolf-Klein, G. (2009). Mandatory cognitive screening in hospitalized elderly people: are we missing the diagnosis? Journal of the American Geriatrics Society 57: 746-747. Chodosh, J., Petitti, D. B., Elliott, M., Hays, R. D., Crooks, V. C., Reuben, D. B., Buckwalter, J. G. and Wenger, N. (2004). Physician recognition of cognitive impairment: evaluating the need for improvement. Journal of the American Geriatrics Society 52: 1051-1059. Corey-Bloom, J., L. J. Thal, L., Galasko, D., Folstein, M., Drachman, D., Raskind, M. and Lanska, D. (1995). Diagnosis and evaluation of dementia. Neurology 45: 211-8. Fillit, H., Geldmacher, D., Welter, R., Maslow, K. and Fraser, M. (2002). Optimizing coding and reimbursement to improve management of Alzheimer s disease and related dementias. Journal of the American Geriatrics Society 50: 1871-8. Folstein, M. (2007). Improving dementia assessment by reducing sample heterogeneity. International Psychogeriatrics 19: 383-9. Greco, A., Cascavilla, L., Paris, F., Errico, M., Orsitto, G., D Alessandro, V., Placentino, G., Franceschi, M., Seripa, D., Vendemiale, G. and Pilotto, A. (2005). Undercoding of Alzheimer s disease and related dementias in hospitalized elderly patients in Italy. American journal of Alzheimer s disease and other dementias 2:, 167-70. Gutterman, E., Markowitz, J., Lewis, B. and Fillit, H. (1999). Cost of Alzheimer s disease and related dementia in managed-medicare. Journal of the American Geriatric Society 47: 1065-71. Hill, J., Futterman, R., Duttagupta, S., Mastey, V., Lloyd, J. and Fillit, H. (2002). Alzheimer s disease and related dementias increase costs of comorbidities in managed Medicare. Neurology 58: 62-70. HIMAA (2006). Introduction to coding with ICD-10-AM, ACHI and ACS. North Ryde, NSW, Health Information Management Association of Australia. Iliffe, S., Haines, A., Gallivan, S., Booroff, A., Goldenberg, E. and Morgan, P. (1991). Assessment of elderly people in general practice: social circumstances and mental state. British Journal of General Practice 41: 9-12. Jorm, A. and Mackinnon, A. (1997). Psychogeriatric assessment scales: user s guide. (Third edition). Mental Health Research Institute. Available at: http://www.biostats.com.au/pas/ Download.html (accessed 17 September 2009). Kinosian, B., Stallard, E., Lee, J., Woodbury, M., Zbrozek, A. and Glick, H. (2000). Predicting 10-year care requirements for older people with suspected Alzheimer s disease. Journal of the American Geriatric Society 48: 631-8. Kukull, W., Higdon, R., Bowen, J., McCormick, W., Teri, L., Schellenberg, G., van Belle, G., Jolley, L. and Larson, E. (2002). Dementia and Alzheimer disease incidence: a prospective cohort study. Archives of neurology 59: 1737-46. Laurila, J., Pitkala, K., Strandberg, T. and Tilvis, R. (2004). Detection and documentation of dementia and delirium in acute geriatric wards. General hospital psychiatry 26: 31-5. McCormick, W., Kukull, W., van Belle, G., Bowen, J., Teri, L. and Larson, E. (1994). Symptom patterns and comorbidity in the early stages of Alzheimer s disease. Journal of the American Geriatric Society 42: 517-21. NCCH (2008). Australian Coding Standards for ICD-10-AM and ACHI. Sydney, Commonwealth of Australia. O Bryant, S., Humphreys, J., Smith, G., Ivnik, R., Graff-Radford, N., Petersen, R. and Lucas, J. (2008). Detecting Dementia with the Mini-Mental State Examination (MMSE) in highly educated individuals. Archives of Neurology 65: 963 7. Park, M., Delaney, C., Maas, M. and Reed, D. (2004). Using a Nursing Minimum Data Set with older patients with dementia in an acute care setting. Journal of advanced nursing 47: 329-39. RACGP (2006). Cornell scale for depression in dementia. In: Medical care of older persons in aged care facilities (Fourth edition). Melbourne, Royal Australian College of General Practitioners. Available at: http://www.racgp.org.au/silverbookonline/4-5.asp Reid, B., Allen, C. and McIntosh, J. (2005). Investigation of leukaemia and lymphoma AR-DRGs at a Sydney teaching hospital. Health Information Management Journal 34: 34-39. Taylor, D. and Sloan, F. (2000). How much do persons with Alzheimer s disease cost Medicare? Journal of the American Geriatric Society. 48: 639-46. 10 HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE)

Wancata, J., Windhaber, J., Krautgartner, M. and Alexandrowicz, R. (2003). The consequences of non-cognitive symptoms of dementia in medical hospital departments. International Journal of Psychiatry in Medicine 33: 257-71. Zilkens, R., Spilsbury, K., Bruce, D. and Semmens, J. (2009). Linkage of hospital and death records increased identification of dementia cases and death rate estimates. Neuroepidemiology 32: 61-9. Corresponding author: Elizabeth Cummings RN, RM, BA, BIS(Hons), PhD Wicking Dementia Research Centre and ehealth Services Research Group School of Computing and Information Systems University of Tasmania, Private Bag 87 Tel: +61 3 6226 7295 email: Elizabeth.Cummings@utas.edu.au Roxanne Maher BSC(HIM) Research Nurse AusLong Study, Menzies Research Institute University of Tasmania, Private Bag 23 Tel: +61 3 6226 7713 email: Roxanne.Maher@utas.edu.au Christine Stirling BN, MPA, PhD Senior Lecturer, School of Nursing and Midwifery University of Tasmania, Private Bag 135, Tel: +61 3 6226 4678 email: Christine.Stirling@utas.edu.au Andrew Robinson Dip App Sc (Nurs), MNSc, PhD Professor of Aged Care Nursing, School of Nursing and Midwifery Co-Director, Wicking Dementia Research and Education Centre University of Tasmania, Private Bag 121 Tel: +61 3 6226 4735 email: Andrew.Robinson@utas.edu.au Paul Turner BA(Hons), MSC, PhD Senior Research Fellow and Director ehealth Services Research Group School of Computing and Information Systems University of Tasmania, Private Bag 87 Tel: +61 3 62266240 email: Paul.Turner@utas.edu.au Christopher Morris Showell BAppSci(MLS) Research Fellow, ehealth Services Research Group School of Computing and Information Systems University of Tasmania, Private Bag 87 Tel: +61 3 62267200 email: chris.showell@utas.edu.au Toby Croft BSc (Hons), PhD Manager of Psychological Services Royal Hobart Hospital, Liverpool Street Hobart TAS 7000 AUSTRALIA Tel: +61 3 6222 7840 email: Toby.Croft@dhhs.tas.gov.au Jane Tolman BA, Dip Ed, MEd, BSc, MBBS, FRACP Director of Aged Care Royal Hobart Hospital, Liverpool Street Hobart TAS 7000 AUSTRALIA Tel: +61 3 6222 7893 email: Jane.Tolman@dhhs.tas.gov.au James Vickers BSc (Hons), PhD, DSc Wicking Dementia Research Centre, University of Tasmania, Private Bag 34, Tel: +61 3 6226 2679 email: James.Vickers@utas.edu.au HEALTH INFORMATION MANAGEMENT JOURNAL Vol 40 No 3 2011 ISSN 1833-3583 (PRINT) ISSN 1833-3575 (ONLINE) 11