Yes, the abstract. One sentence cites a result suggesting that 8895 IU/d is necessary to reach >= 50nmol/L and the next another that 6201 IU/d is necessary to reach 75 nmol/L. Clearly these aren't compatible to even 1 significant figure. Now, of course this is a matter of accuracy and not of precision, but do you really believe someone's 4 sig figs of precision is at all justified when they disagree with another person's 4 sig figs to such an extent? It's possible, I guess! But I'm betting if you look at the original study you only have 1 or maybe 2 significant figures that are actually justified.
Look further at the paper where the 8895 number comes from [1]. That's the only number reported to more than 2 sig figs in the paper and comes in this context [emphasis mine]: "It also estimated that 8895 IU of vitamin D per day may be needed to accomplish that 97.5% of individuals achieve serum 25(OH)D values of 50 nmol/L or more. As this dose is far beyond the range of studied doses, caution is warranted when interpreting this estimate. Regardless, the very high estimate illustrates that the dose is well in excess of the current RDA of 600 IU per day and the tolerable upper intake of 4000 IU per day [1]." The two figures don't even show half the range needed to see this value and make it pretty clear that this is a laughable amount of precision to report. To make this clear: the value reported is the intercept of the dashed red line with the 50nmol/L line in the second figure -- which has been extrapolated out to a value almost three times as large as the highest dose used in any experiment. And the extrapolation is done on a curve that is getting very close to level at that point so even a minor error in dose response would cause a large error in reported dose. Frankly, after looking at this, I don't think the originally posted article should have quoted this number in its abstract - the point of this paper seems to be to show that the standard value of 600 IU/day is unsupported form the data, not to suggest that their 8895 IU/day is well supported.
Consider the most well-studied nutritional number we ever see: the daily caloric intake! It's never given more than 2 sig figs and often just 1! And this study is making the argument that the daily recommendations for this are off by a factor of 10x. It'll take a lot to convince me that any number in nutrition can be reported to 4 significant figures, and even then I expect that it would be useless since inter-person variation is going to cause at least a 1% difference (probably more like 25% differences like calories just for men/women differences).
[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4210929/