This analysis takes the final observation for each patient and uses it as that patient's endpoint in the analysis. For example, in a 12 month trial in acute schizophrenia, a patient who withdraws at month 7 due to side effects will have their month 7 value included in the analysis of the data.
In one sense this approach has clinical appeal. The final value provided by the patient who withdrew at month 7 is a valid measure of how successful we have been in treating this patient with the assigned treatment and so should be part of the overall evaluation of that treatment. In some circumstances, however, this argument breaks down. For example, if there is an underlying worsening trend in disease severity then patients who withdraw early will tend to provide better outcomes than those who withdraw later on in the treatment period. If one treatment has more early dropouts than the other, possibly because of side effects say, then there will be bias caused by the use of LOCF. Multiple sclerosis and Alzheimer's disease are settings where this could apply. The opposite will of course be true in cases where the underlying trend is one of improvement; depression would be one such therapeutic area. These scenarios emphasise the earlier point that there is no universally valid way to deal with missing data.
Was this article helpful?