Dementia: Predictors of diagnostic accuracy and the contribution of diagnostic recommendations

,

Original Research

Dementia: Predictors of diagnostic accuracy and the contribution of diagnostic recommendations

J Fam Pract. 2002 August;51(8):693-699

PDF Download

References

1. Eefsting JA, Boersma F, Van den Brink W, Van Tilburg W. Differences in prevalence of dementia based on community survey and general practitioner recognition. Psychol Med 1996;26:1223-30.

2. O’Connor DW, Pollitt PA, Hyde JB, Brook CP, Reiss BB, Roth M. Do general practitioners miss dementia in elderly patients? BMJ 1988;297:1107-10.

3. Valcour VG, Masaki KH, Curb JD, Blanchette PL. The detection of dementia in the primary care setting. Arch Intern Med 2000;160:2964-8.

4. Verhey FR, Jolles J, Ponds RW, et al. Diagnosing dementia: a comparison between a monodisciplinary and a multidisciplinary approach. J Neuropsychiatry Clin Neurosci 1993;5:78-85.

5. Wind A, Muskens J, De Bruyne G, et al. NHG standaard M21 Dementie [Dutch dementia guideline]. Utrecht, The Netherlands: Nederlands Huisartsen Genootschap; 1999.

6. Eccles M, Clarke J, Livingstone M, Freemantle N, Mason J. North of England evidence based guidelines development project: guideline for primary care management of dementia. BMJ 1998;317:802-8.

7. Rose VL. Consensus statement focuses on diagnosis and treatment of Alzheimer’s disease and related disorders in primary care. Am Fam Physician 1998;57:1431-2.

8. Wind AW, Van Staveren G, Schellevis FG, et al. The validity of the judgement of general practitioners on dementia. Int J Geriatr Psychiatry 1994;9:543-9.

9. Cooper B, Bickel H, Schäufele M. The ability of general practitioners to detect dementia and cognitive impairment in their elderly patients: a study in Mannheim. Int J Geriatr Psychiatry 1992;7:591-8.

10. Pond CD, Mant A, Kehoe L, Hewitt H, Brodaty H. General practitioner diagnosis of depression and dementia in the elderly: can academic detailing make a difference? Fam Pract 1994;11:141-7.

11. O’Connor DW, Fertig A, Grande MJ, et al. Dementia in general practice: the practical consequences of a more positive approach to diagnosis. Br J Gen Pract 1993;43:185-8.

12. van Hout H, Vernooij-Dassen M, Bakker K, Blom M, Grol R. General practitioners on dementia: tasks, practices and obstacles. Patient Educ Couns 2000;39:219-25.

13. Colerick EJ, George LK. Predictors of institutionalization among caregivers of patients with Alzheimer’s disease. J Am Geriatr Soc 1986;34:493-8.

14. Vernooij-Dassen MJFJ, Persoon JMG, Felling AJA. Predictors of sense of competence in caregivers of demented persons. Soc Sci Med 1996;43:41-9.

15. McCurry S, Teri L. Advance planning for dementia caregivers. J Am Geriatr Soc 1997;45:1102-3.

16. Droës R. Psychosocial treatment for demented patients: overview of methods and effects. In: Miesen B, Jones G, Eds. Care-Giving in Dementia, Research and Applications. London: Routledge; 1997;127-48.

17. Rösler M, Anand R, Cicin-Sain A, et al. Efficacy and safety of rivastigmine in patients with Alzheimer’s disease: international randomised controlled trial. BMJ 1999;318:633-8.

18. Rogers SL, Farlow MR, Doody RS, Mohs R, Friedhoff LT. A 24-week, double-blind, placebo-controlled trial of donepezil in patients with Alzheimer’s disease. Donepezil Study Group. Neurology 1998;50:136-45.

19. De Bruyne GM, Meyboom-de Jong B, Muskens JB, et al. De NHG-standaard dementiesyndroom [Guideline on dementia of the Dutch College of General Practitioners]. Huisarts & Wetenschap 1991;34:598-607.

20. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed rev. Washington, DC: American Psychiatric Association; 1987.

21. Beck C, Cody M, Souder E, Zhang M, Small GW. Dementia diagnostic guidelines: methodologies, results, and implementation costs. J Am Geriatr Soc 2000;48:1195-203.

22. Van Hout H, Vernooij-Dassen M, Poels P, Hoefnagels W, Grol R. Applicability of diagnostic recommendations on dementia in family practice. Int J Qual Health Care 2001;13:127-33.

23. Roth M, Tym E, Mountjoy CQ, et al. CAMDEX, a standardised instrument for the diagnosis of mental disorder in the elderly with special reference to the early detection of dementia. Br J Psychiatry 1986;149:698-709.

24. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.

25. Mendez MF, Mastri AR, Sung JH, Frey WH, 2nd. Clinically diagnosed Alzheimer disease: neuropathologic findings in 650 cases. Alzheimer Dis Assoc Disord 1992;6:35-43.

26. Rasmusson DX, Brandt J, Steele C, Hedreen JC, Troncoso JC, Folstein MF. Accuracy of clinical diagnosis of Alzheimer disease and clinical features of patients with non-Alzheimer disease neuropathology. Alzheimer Dis Assoc Disord 1996;10:180-8.

27. Hosmer D, Lemeshow S. Applied Logistic Regression. New York: Wiley; 1989.

28. Harrell FE, Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361-87.

29. Moons KG, Stijnen T, Michel BC, et al. Application of treatment thresholds to diagnostic-test evaluation: an alternative to the comparison of areas under receiver operating characteristic curves. Med Decis Making 1997;17:447-54.

30. Moons KG, van Es GA, Michel BC, Buller HR, Habbema JD, Grobbee DE. Redundancy of single diagnostic test evaluation. Epidemiology 1999;10:276-81.

31. Hanley J, McNeil B. The meaning and use of the area under the receiver operating characteristic (ROC) curve. Radiology 1982;143:29-36.

32. Murphy JM, Berwick DM, Weinstein MC, Borus JF, Budman SH, Klerman GL. Performance of screening and diagnostic tests. Application of receiver operating characteristic analysis. Arch Gen Psychiatry 1987;44:550-5.

33. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148:839-43.

34. Harmsen J, Hingstman L. Cijfers uit de registratie van huisartsen, peiling 1996 [Numbers of the registration of General Practitioners in the Netherlands]. Utrecht, The Netherlands: NIVEL; 1996 November.

35. Schmidt HG, Norman GR, Boshuizen HPA. A cognitive perspective on medical expertise: theory and implications. Acad Med 1990;65:611-21.

36. Custers EJ, Boshuizen HP, Schmidt HG. The influence of medical expertise, case typicality, and illness script component on case processing and disease probability estimates. Mem Cognit 1996;24:384-99.

37. Erkinjuntti T, Ostbye T, Steenhuis R, Hachinski V. The effect of different diagnostic criteria on the prevalence of dementia. N Engl J Med 1997;337:1667-74.

38. Van den Hoogen H, Huygen F, Schellekens J, et al. Morbidity figures from general practice, data from four general practices 1978–1982. Nijmegen, The Netherlands: Nijmegen University Department of General Practice; 1985.

TABLE 4

Multivariate associations between predictors of accurate dementia diagnoses by family physicians and the reference standard memory clinic team diagnosis

	Adjusted odds ratio (95% CI)	P
ADL dependency, score1–4	5.35 (2.00–13.9)	.0212
Years since symptoms started	1.84 (1.08–3.14)	.026
Somatic comorbidity present	0.48 (0.25–0.89)	.006
n = 93.
ADL, activities of daily living (1 = independent to 4 = fully dependent); CI = confidence interval.

Predictors of presence/absence of dementia

The univariate analyses revealed that informant availability, years of education, sex of the family physician, number of cognitive symptoms, ADL dependency, somatic comorbidity, blood abnormality, behavioral changes, and duration and severity of the symptoms were associated with the presence or absence of dementia (Table 3). Neither the number of applied recommendations nor the use of the MMSE was related to the presence or absence of dementia.

The multivariate logistic regression model (Table 5) revealed that ADL dependency (OR = 5.28, P = .001) and time since the first symptoms started (OR per year =1.84, P = .026) independently predicted the presence of dementia, whereas somatic comorbidity predicted the absence of dementia (OR = 0.48, P = .021).

The ROC area of the model with all 9 significant univariate variables from Table 4 was 0.86 (95% confidence interval [CI], 0.77–0.94). Reduction of the model to the 3 significant multivariate variables (ADL, time since the first symptoms were noticed, and somatic comorbidity) resulted in an ROC area of 0.79 (95% CI, 0.70–0.88). The family physicians’ global diagnosis had an ROC area of 0.74 (95% CI, 0.63–0.85). Finally, the ROC area of the expected DSM-III-R diagnosis was only 0.50 (95% CI, 0.37–0.62).

Discussion

The family physicians’ diagnosis based on their assessment of all available information was reasonably accurate. Formal DSM-III-R diagnoses, derived by integrating the recorded symptoms, resulted in poor accuracy. Degree of ADL functioning, the duration of symptoms, and the presence of somatic comorbidity independently contributed to the prediction of presence or absence of dementia. Neither the number of diagnostic recommendations made nor the use of the MMSE added to the family physicians’ accuracy. The latter is rather disappointing news for makers of all 3 dementia guidelines. Many of the current diagnostic recommendations in the dementia guidelines are based at best on lower quality evidence from observational studies. Trials are lacking in which diagnostic interventions by family physicians are tested against usual diagnostics.

It was remarkable that neither the core symptoms of dementia (cognitive impairment) nor the DSM criteria added to the diagnostic accuracy. Our findings raise the question of how family physicians made their diagnoses. In this respect, the concept of illness scripts (also called pattern recognition) may be helpful to understand how clinicians make diagnostic decisions.³⁵ According to this concept, clinicians base their decisions more on accumulated clinical patient pictures during their medical career or so-called illness scripts than on medical-deductive reasoning, which is taught in medical schools and offered in evidence-based guidelines.³⁶ The illness scripts family physicians have of suspected dementia patients may be triggered more by ADL malfunctioning than by cognitive dysfunction.

When the time effort per patient of the family physicians and the memory clinic is compared, the family physicians performed well. The family physicians invested on average 3.5 consultations per patient. Multiplied by the average consultation time of 10 minutes in Dutch family practices, this yields a total consultation time of 35 minutes per patient to arrive at a diagnosis. This time should be compared with the 4.5 hours needed per patient for high-tech diagnostic assessment at the memory clinic. Taking the limited time and low-tech approach into account, it must be concluded that the family physicians did a fair job in diagnosing suspected patients.

We did not confirm the hypothesis that continuity of care enhanced diagnostic accuracy. In contrast to the finding of Eefsting and colleagues,¹ the number of contacts with a patient showed a trend toward a negative association with diagnostic accuracy. A possible explanation is that only contacts related to the evaluation of dementia were recorded in our study; a larger number of contacts may therefore reflect a family physicians’ diagnostic difficulty or uncertainty rather than better continuity of care.

Strengths of our study include the detailed information from the family physician assessments, the primary care setting, and the use of a memory clinic as the reference standard. A limitation was that the diagnostic criteria of the family physicians’ guideline were based on the DSM-III-R, whereas the memory clinic used the DSM-IV criteria.^20,24 Nevertheless, this difference may have accounted for the diagnostic variation only to a small degree.³⁷ Second, the finding that additional value was not found for use of the MMSE or for a more frequent use of recommendations might be explained by the fact that the family physicians in this study already applied the recommendations on a large scale. Higher use, therefore, did not add much to the diagnostic accuracy of the family physicians. Third, concerning the importance of the availability of an informant, the kind of information the family physicians received from these informants remains unclear, as the content of the informant interview was neither standardized nor registered. Fourth, few patients per family physician were included. The numbers of included patients per family physician, however, are only a little below the incidence figures for dementia in the Netherlands.³⁸ Anecdotal reactions from the participating family physicians revealed that some patients were not included because they did not want medical interference and some were unable to travel to the memory clinic. Finally, we modeled various predictors to estimate the presence or absence of dementia and derived a reduced model. In this modeling we did not adjust for overfitting and did not prospectively validate the model. This tactic was necessary given the relatively small number of subjects in the study.²⁸ Therefore, before our reduced prediction model can be used in medical practice it should first be validated on new patients.