Objectives. The aims of the article are, firstly, to analyze the effect of different methods of post-stroke depression assessment on estimated prevalence rates of this condition, and secondly, to consider other possible factors contributing to significant discrepancies in the available epidemiological data.
Method. Data on prevalence of post-stroke depression (PSD) reported in 44 English-language papers indexed by the PubMed data-base were analyzed. The data were categorized with regard both to the type of diagnostic instruments used (clinical classification systems, diagnostic scales or self-report questionnaires) and populations studied (general population, inpatients, and discharged patients). The reported prevalence rates in these populations were compared controlling for the diagnostic instrument used.
Results. There were statistically significant differences in the PSD prevalence rates in the general population depending on the type of diagnostic tools used, ranging from about 14% (in terms of clinical scales), through almost 24% (by diagnostic criteria of classification systems), to almost 32% (self-report questionnaires). As regards the inpatient population, the highest PSD prevalence rates (over 39%),obtained using classification systems, were significantly higher than these estimated with the use of the remaining two types of diagnostic tools (both indicating slightly over 31%). In the population of discharged patients the lowest prevalence of PSD (over 5%) was indicated by classification systems. This rate was significantly lower than the concordant estimates of PSD prevalence (slightly over 31%) obtained from the other two diagnostic tools.
Conclusions. Self-report questionnaires, clinical scales and classification systems differ in their accuracy of the PSD assessment. In each of the three populations analyzed significant measurement differences were found between the three categories of diagnostic tools, which may result in the discrepancies in reported PSD prevalence rates. Thus, it seems desirable to assess the diagnostic accuracy of particular instruments, as well as the effect of time from stroke on PSD onset.