The primary variables of interest, however, were platelet count and hemoglobin levels, either lagged by one study visit or as an immediate marker.
Compared to a reference group whose platelets dropped by no more than 21,000 per microliter from baseline, those with a greater decline were at twice the risk of a diagnosis of dementia at the next study visit. The adjusted hazard ratio was 2.39, with a 95% confidence interval from 1.14 to 5.02, which was significant at P=0.02.
However, hemoglobin levels - measured at the visit before a diagnosis of dementia - were not significantly associated with increased risk.
On the other hand, platelet levels, measured at the same time as the dementia diagnosis, were no longer significantly associated with increased risk, while the opposite was true for hemoglobin.
Compared to a reference group with normal levels, low hemoglobin - defined as 2.7 to 12.6 grams per deciliter of blood - was significantly associated with dementia. The adjusted hazard ratio was 2.60, with a 95% confidence interval from 1.31 to 5.17, which was significant at P=0 .006.
Interestingly, the relationship between platelet level and dementia risk appeared to change over time, the researchers found. Specifically:
- During the first two years of follow-up, a platelet decline at the visit immediately before a diagnosis of dementia was associated with an even greater risk of dementia, compared to those whose platelet counts were stable. The adjusted hazard ratio was 6.76 with a 95% confidence interval from 2.36 to 19.41, which was significant at P<0.001.
- In contrast, the association was no longer evident if the decline occurred after two years of follow-up.
While the study was prospective, the researchers noted that the results might have been biased by the need to exclude patients who developed HIV by the first follow-up visit after baseline.