Reliability in diagnosis - PowerPoint Presentation

Slide1

Reliability in diagnosis

Reliability of diagnosis

= the extent to which clinicians agree on the same diagnosis for the same patient

.

The extent to which the same diagnosis is given in different cultures

Reliability in diagnosis

Ward et al (1962) – low agreement between 2 psychiatrists because of inconsistent information provided by patient, inconsistent interpretation by the psychiatrist and inadequacy of the classification tool.

Beck (1954) – low inter-

rater

reliability, same diagnosis given by 2 psychiatrists in about 50% of cases (testing older version of DSM)

Some symptoms have high level of overlap of symptoms (e.g. PTSD) which makes it hard to give reliable diagnosis

Reliability in diagnosis

Patient factors:

inconsistency can be caused by the patient giving inaccurate information due to poor memory, denial, shame, the symptoms of their disorder (e.g. disorganised speech, manipulative tendencies)

Clinician factors:

inconsistency can be caused by the clinician focusing on one particular type of symptoms, using their subjective judgement (based on their background and training), implicit bias about a certain type of people (e.g. culture or gender)

Reliability in ICD/DSM

Goldstein (1988) found inter-

rater

reliability in the DSM-III for patients with Schizophrenia

Brown et al (2001) found overall reliability in DSM-IV but some issues with boundaries – if a patient fell on the boundary of 2 similar disorders it was hard to distinguish which one they had.

Jakobsen

et al (2005) concluded ICD-10 gave a reliable diagnosis of schizophrenia, as it showed good agreement with a second diagnosis using a different tool.

Hiller et al (1992) compared the ICD-10 with the DSM-III-R and found the ICD-10 gave higher reliability for all disorders studied except bipolar disorder.

Cheniaux

et al (2009) found schizophrenia was diagnosed more frequently when the ICD was used instead of the DSM which suggests a lack of reliability between the systems.

Reliability figures in the studies are around 0.50 agreement, which still leaves a large amount of disagreement to consider.

Validity in diagnosis

Validity in diagnosis =

must genuinely reflect the underlying disorder. Can be established by having concurrent validity (using both DSM and ICD to diagnose), aetiological validity (looking at the causes) and predictive validity (checking the diagnosis against the outcome

).

Validity can be affected by the interchange between the clinician and the patient. The clinician may have

implicit bias

(e.g. more likely to diagnose a woman with depression) which could give a subjective diagnosis.

The validity issues can be caused by the clinician’s training and perception of presenting symptoms (

Aboraya

et al 2006)

Validity in the DSM/ICD

There is a lack of construct validity in the DSM, as it

operationalises

mental disorders into a list of symptoms, therefore losing out on the bigger picture of the disorder – the list of symptoms may not be representative enough of the disorder.

Kim-Cohen et al (2005) found good validity in the DSM-IV if multiple data sources were used (mother’s responses, teachers responses, observational data, comparisons with other children, all led to a valid diagnosis of conduct disorder in children)

Splitting a mental health disorder into a list of symptoms is reductionist, and a holistic approach might be more valid

.

Validity in the DSM/ICD

The use of questionnaires and interviews used to study the validity and reliability of the DSM can lack validity, especially if their purpose is outlined beforehand

The DSM is developed in the USA and there are some cultural issues in diagnosing mental disorders.

Luhrmann

et al (2015) found the hearing voices in the USA is seen as negative, but in Ghana and India this is seen as a positive experience

Burnham et al (1987) found more auditory hallucinations were reported by patients who were Mexican born Americans than those who were non-Mexican born Americans.

Chandrasena

(1986) reported more incidences of catatonia in Sri Lanka (21% compared to 5% among British white people)

Pihlajamaa

et al (2008) found that generally when the ICD is used to diagnose schizophrenia the diagnosis matches a diagnosis given by a different system, showing concurrent validity.