I read with interest the article by Bonell-Pascual et al (Reference Bonell-Pascual, Huline-Dickens and Hollins1999) and would agree that recognition of, and service provision for, the bereaved adult learning disability population is markedly deficient. However, their conclusion that learning disability is a significant predictor of future mental health problems following bereavement is not substantiated and should be interpreted with caution in view of the flawed research methodology.

The original cohort reported short-term psychopathological and behavioural outcomes following bereavement (Reference Hollins and EsterhuyzenHollins & Esterhuyzen, 1997). This case-controlled study found an increase in aberrant behaviours and psychopathology in those bereaved compared with controls. Bonell-Pascual et al aimed to investigate whether the same outcomes had changed over a longer term (six to eight years). Unfortunately, the control cohort was not followed up, thereby making meaningful inferences impossible, as confounding life events are not controlled for. Also, more than half the study population had additional medical disorders of likely prognostic significance. The authors recognise the limitations of their small sample size, but this is further compounded by incomplete follow-up of the original cohort, with greater than 15% of the original cohort excluded. Furthermore, follow-up data are missing from the analysis of psychopathology and aberrant behaviour.

In the classification of psychopathology, each sub-scale shows varied changes with no interpretable trends over time. Psychopathology identified in the original 1997 study had resolved in over 70%, although a few new cases were identified, especially in the adjustment and anxiety disorder sub-categories. However, we cannot attribute these new cases to underlying learning disability as confounding medical and life events may have played a significant role. Aberrant behaviour patterns were reported as showing a mean increase in each sub-scale over time. The individual change in aberrant behaviour patterns was, however, in both directions in all sub-scales (i.e. an improvement and deterioration in behaviour). The mean change quoted by the authors in such a small study sample with wide ranges of behaviour is, therefore, unreliable. Similarly, the one statistically ‘significant’ result, showing inappropriate speech to be more common, should also be treated with caution and taken in isolation is unlikely to be of clinical significance. Furthermore, the authors allude to problems with validation of the behavioural and psychopathological measures used in the significant proportion of the study population who were taking antipsychotic and antidepressant medication.

It is now recognised that people with learning disability do understand the permanence of death and consequently grieve (Reference Harper and WadsworthHarper & Wadsworth, 1993; Reference ReadRead, 1996). Bonell-Pascual et al highlight the need to better understand the general and specific needs of this vulnerable group of people and provide some useful insight into effective interventions. Future studies should be aimed at addressing these issues, be carried out prospectively and have adequate control groups to allow safe and meaningful conclusions to be drawn.