Year : 2014  |  Volume : 56  |  Issue : 2  |  Page : 117--120

Diagnostic and Statistical Manual-5 and dementia: Fine print, finer points

KS Jacob 
 Professor of Psychiatry, Christian Medical College, Vellore, Tamil Nadu, India

Correspondence Address:
Dr. K S Jacob
Professor of Psychiatry, Christian Medical College, Vellore - 632 002, Tamil Nadu

How to cite this article:
Jacob K S. Diagnostic and Statistical Manual-5 and dementia: Fine print, finer points.Indian J Psychiatry 2014;56:117-120

How to cite this URL:
Jacob K S. Diagnostic and Statistical Manual-5 and dementia: Fine print, finer points. Indian J Psychiatry [serial online] 2014 [cited 2020 Aug 11 ];56:117-120
Available from:

Full Text


The Diagnostic and Statistical Manual (DSM) revolution has marched on with the unveiling of its most recent achievement, the DSM-5. [1] However, many questions and controversies raised [2] call for introspection into its many diagnostic categories and their implementation in practice. The limitations of phenomenological diagnoses and clinical classifications and the complexity of the diagnostic process, frequently misunderstood, demand review. This editorial attempts to appraise clinical reality, diagnostic guidelines, and medical practice in relation to neurocognitive disorders (NCD).

 Changes in DSM-5

The DSM-5 has made many changes to the section on cognitive disorders. [1] It has expanded cognitive domains, which now include complex attention, executive function, learning and memory, language, perceptual-motor, and social cognition. It changed the name dementia to major NCD, which now subsumes dementia and amnestic disorder. It introduced the category of mild NCD, for mild cognitive impairment. Common etiological diagnoses, which result in neurocognitive presentations, have operational diagnostic criteria.

Clinical information and the results of neuropsychological tests are now required for a definitive diagnosis of NCD. DSM-5 demands comparison of the person's individual performance on neuropsychological tests with population norms, adjusted for age, education, and cultural background. It provides for "bedside" cognitive tests, for objective assessments.

Fine print, finer points

Diagnostic and laboratory tests support the operational criteria and categories for cognitive disorders in DSM-5. The criteria are more rigorous and are on firmer etiopathological evidence base than most other diagnoses in the manual. However, the DSM-5 acknowledges, in the text, many caveats and cautionary statements, which call for a nuanced understanding and cautious interpretation.

Clinical dimensions

The DSM-5 clearly acknowledges that cognitive decline is more appropriately mapped on a dimension from deficits commonly seen in older people without disability, to those with mild cognitive impairment to frank and severe cognitive deterioration. Despite its mathematical and clinical convenience, dichotomous demarcations often misrepresent clinical reality, which often lie on a spectrum, [3] leading to errors. The continuum of cognition and cognitive decline makes categorical diagnosis complex, making the placement of a diagnostic threshold, in stone, difficult. Nevertheless, clinical medicine and psychiatry have preferred diagnostic categories, even while identifying and managing clinical phenomena, which usually lie on a continuum with a gradual change from normal to abnormal/pathological and without definitive threshold. DSM-5 does acknowledge that the current categorization of NCD, into major and mild, as arbitrary.

Surrogate indicators of pathology

The evidence generated by medical procedures contributes different weights to diagnosis. Certain procedures, such as biopsies, often produce definitive evidence. Other procedures, such as the brain imaging in patients with significant cognitive decline, provide contributory evidence. When combined with a history, clinical examination and neuropsychological data, the results of the test can lead to a diagnostic decision.

Nevertheless, the scenario is complex, particularly in the context of dementia. The inability to biopsy brains of people with cognitive impairment forces us to employ surrogate clinical markers for diagnosis. However, the use of surrogate markers is complicated by the limited correlation between the clinical continuum and the spectrum of pathology. The Nun study clearly documents the presence of pathological changes even among those who did not demonstrate clinical evidence of dementia. [4] Consequently, the disconnect or rather the lack of lock step, between pathological and clinical dimensions make diagnostic decision-making using surrogate clinical markers difficult.

Minor criterion variation and prevalence

Studies have documented significant variations in prevalence of dementia within the same population based on minor differences in definitions of the condition. [5] The changes in the operational framework, the expansion of the cognitive domains, and the adjustments in operational criteria in the DSM-5 will result in changes in identification rates, population incidence, and prevalence.

Clinical heterogeneity

Classical presentations of common diseases are diagnosed by pattern recognition. However, the equal status given to the many different and diverse cognitive domains in the DSM-5 also means the need to recognize numerous and varied clinical patterns. The polythetic format of the DSM system implies a vast number of permutations and combinations, which can reach the suggested diagnostic threshold. The assortment of patterns of cognitive decline in the general population demand caution in the application of simplistic and universal diagnostic rules. In addition, inductive logic (Bayesian), employed in medical diagnosis, does not have the same degree of certainty, as it moves from a set of specific facts to a general conclusion.

Population norms

DSM-5 recommends the use of neuropsychological testing or in its absence, other quantified clinical assessment to document impairment in cognitive performance. It suggests a comparison with population norms appropriate to the patient's age, educational attainment, and cultural background. It recommends a diagnosis of major NCD for performance more than two standard deviations below the population norm or less than the third percentile. The recommended norms for mild NCD are fixed between the 3 rd and 16 th percentile or between 1-2 standard deviations. However, DSM-5 also acknowledges that neuropsychological testing is not available in all settings, that neuropsychological thresholds are sensitive to the specific test(s) and norms employed, as well as to test conditions, sensory limitations, and intercurrent illness. It also suggests specific "bedside" assessments to objectively document current performance in settings where such testing is unavailable or not feasible.

Complicating the issues is the fact the person's age, their level of education, and cultural background influence the results of cognitive testing. Education and culture fair tests are more in the ideal realm than in actual practice. The DSM-5 clearly acknowledges the impact of education and culture on the diagnosis of dementia. It advices caution in interpretation of diagnostic data under conditions of extreme education-illiteracy and very high standard of learning. Paper and pencil tests in older people who have not received a formal education are fraught with difficulty. Culture-specific tests are not being available for many cultures. Educational and cultural diversity within populations mean the absence of national, regional, and culture-specific norms. While the DSM-5 suggestion of comparison with population norms is ideal, they will be followed in the breach particularly in clinical practice.

Cross-sectional evaluations of decline

The DSM-5 highlights that the fact that the person's cognitive performance will necessarily have to be interpreted in the light of past performance, to document decline. It suggests that in the absence of such information, the diagnosis can be inferred based on appropriate norms, keeping in mind the individual's educational attainments, occupational achievements and past cognitive functioning. Acquired cognitive decline implies a reduction in ability and function relative to a baseline. Consequently, cross-sectional assessments, without regard to prior ability and functioning, may not reflect change required for diagnosis. The "normal" performance of people with a history of very high levels of cognitive function, although representing marked decline, would not be identified as abnormal, while those with a history of low function will be diagnosed to have pathology on cross-sectional tests. DSM-5 admits that documenting cognitive performance allows for subsequent comparisons.

Statistics of agreement and of prediction

A surrogate diagnostic test is judged by its agreement with the gold standard. Many tests have reasonable indices or averages, which reflect the number of people with disease who are identified by the test (sensitivity) and the number of people without disease who are test negative (specificity). However, the predictive value of a test, when applied in practice, is dependent on the prevalence of the condition in the population tested. [3],[6] Tests used in a group of individuals with low prevalence of the condition to be detected would produce high false positive rates. [7] Consequently, the use of the same dementia criteria in varied settings with their differing prevalence (e.g., specialist, primary care, and community) have a different probability to accurately reflect clinical reality. The DSM system has argued for universalization of psychiatric diagnosis across location and settings. Nevertheless, the impact of prevalence of the condition in the population tested on predictive values strongly suggests caution. This is particularly true in low prevalence situations like primary care and community settings.

Screening conundrum

The multiple cognitive domains and the polythetic formats for diagnosis mean that people who receive such labels are diverse in their cognitive profiles. Clinical heterogeneity, also the norm even within similar etiopathological categories, has major implications for screening to identify cases. Even good diagnostic tests, which use surrogate markers of pathology, are not to be employed indiscriminately in populations with low prevalence. [7] However, most screens for cognitive impairment rarely highlight, [8] let alone discuss their high false-positive rates when employed in situation of low prevalence.

Most current screening instruments do not have an equal scoring system for each of the cognitive domains. Consequently, any threshold set will not do justice to the variation in clinical presentations. The expanded cognitive domain system in the DSM-5 and its new criteria will necessarily mean the need to develop new screening instruments to incorporate these changes. It will also most certainly mean the need for the revalidation of existing screens.

Diagnostic certainty and risk

The degree of diagnostic certainty needed in making clinical decisions is also a function of the degree of risk presented by the therapeutic options. For the use of specific therapy, which is efficacious and has a low level of risk of adverse effects (e.g., the use of vitamin supplementation in old age), few tests are needed because physicians can accept substantial diagnostic uncertainty. On the contrary, in situations where treatment options are less effective and carry greater risk of side-effects (e.g., antipsychotic use in vascular disease), clinicians often need a higher degree of diagnostic certainty. This issue is rarely highlighted in debates on diagnosis and classification.

Health system issues

There are many easy and inexpensive procedures, which are surrogate and substitute for more definitive tests, and are employed to screen for different conditions. Those positive on such screens are subsequently confirmed using an expensive or elaborate tests. However, when this principle is employed at the level of health systems it makes issues complex. The World Health Organization [9] and the 10/66 Dementia Research Group [10] recommend the strategy to employ trained community health workers to identify dementia in low- and middle-income countries (LMIC) and in resource poor settings. They argue that such an approach is cost-effective, utilizes existing resources without increasing their workload. However, studies, which have systematically compared such diagnosis with gold standards, have shown very high false-positive rate and unacceptable predictive values. [7] The low prevalence situations in community preclude the use of routine screens, even those that are useful in situations of high prevalence. Successful and efficient screening strategies require an optimal prevalence of the condition in the population tested. This is achieved only by having a tiered referral system. The recommendation of employing trained health workers without a tiered health care system is the reason for the persistence of very poor identification and treatment rates for mental illness in general and for dementia in particular in LMIC.


A clinician is required to make clear decisions based on an unambiguous estimate of the problem. Patients usually seek and physicians often provide such a definitive diagnosis and this works well in practice. However, often, the clinical picture is often ambiguous, making it difficult to reach a definitive conclusion. In such situations, the possibility of making a mistake is real and is a common professional hazard for all physicians. Rather than accepting the ambiguity of certain clinical situations and explaining this to patients, doctors are often pressured to make definitive decisions in unclear circumstances. Situations, which actually demand a probabilistic inference due to the incomplete and fragmentary nature of the available information, are often discussed in terms of clinical certainty, forcing errors.

There is no such thing as a perfect diagnostic system; improvements made often have a trade-off. Highly sensitive systems overdiagnose conditions, while blunter investigative methods underestimate the risks. The trade-off is essentially between sensitive systems, which give false alarms, and blunt systems, which do not pick up the condition concerned. The diagnostic challenge for physicians is to separate the signal from background noise. Arriving at a medical diagnosis is a complex process requiring clinical skill. The need for clear decisions has to be balanced by an acceptance of the ambiguity of the situation. The challenge is to integrate the science and the art of clinical medicine. Nevertheless, universal criteria and categories are appealing at many levels; they provide certainty in our confusing clinical and research worlds. However, there is a definite need for a contextual understanding of the diagnosis of major and minor NCD. [11]

The DSM-5 while pushing for universal diagnostic criteria and categories recognizes the need for contextualisation of the issues. The complexity of cognitive decline and intricacies of the process of clinical diagnosis add to diagnostic uncertainty. Understanding the issues will help both physicians and patients make the best decisions related to health.


1American Psychiatric Association. diagnostic and statistical manual of mental disorders. 5 th ed. Arlington VA: APA; 2013.
2Insel T. Transforming diagnosis. 29.4.2013. Available from: 29.4.2013 [Last accessed on 2013 Sept 9].
3Feinstein AR. Clinical Epidemiology: The architecture of clinical research. Philadelphia: WB Saunders; 1985.
4Snowdon DA. Aging and Alzheimer′s disease: Lessons from the Nun Study. Gerontologist 1997;37:150-6.
5Jacob KS, Kumar SP, Gayathri K, Abraham S, Prince M. The diagnosis of dementia in the community. Int Psychogeriatr 2007;19:669-78.
6Coggon D, Rose G. Barker DJP. Epidemiology for the uninitiated, fourth edition. [Last accessed on 2013 Sept 9].
7Jacob KS, Senthil Kumar P, Gayathri K, Abraham S, Prince MJ. Can health workers diagnose dementia in the community? Acta Psychiatr Scand 2007;116:125-8.
8Cherbuin N, Anstey KJ, Lipnicki DM. Screening for dementia: A review of self- and informant-assessment instruments. Int Psychogeriatr 2008;20:431-58.
9World Health Organization. The introduction of the mental health component into primary health care. Geneva: WHO; 1990.
10Shaji KS, Arun Kishore NR, Lal KP, Prince M. Revealing a hidden problem. An evaluation of a community dementia case-finding program from the Indian 10/66 dementia research network. Int J Geriatr Psychiatry 2002;17:222-5.
11Jacob KS. Dementia: Toward contextual understanding. Int Psychogeriatr 2012;24:1703-7.