We don't know. Since you asked, here's the comment from one of the more recent meta-analyses to discuss dose in connection with all-cause mortality, Autier 2014:

Results of meta-analyses and pooled analyses consistently showed that supplementation could significantly reduce the risk of all-cause mortality, with relative risks ranging from 0·93 to 0·96 (table 4). Most trials included elderly women and a sizeable proportion of individuals were living in institutions. Decreases in risks of death were not associated with trial duration and baseline 25(OH)D concentration.^13^ Mortality reductions in trials that used doses of 10–20 μg per day of vitamin D seemed greater than were reductions noted with higher doses.^13,14^

• 13. Bjelakovic G, Gluud LL, Nikolova D, et al. "Vitamin D supplementation for prevention of mortality in adults". Cochrane Database Syst Rev 2011; 7: CD007470. [2014 update]
• 14. Rejnmark L, Avenell A, Masud T, et al. "Vitamin D with calcium reduces mortality: patient level pooled analysis of 70,528 patients from eight major vitamin D trials". J Clin Endocrinol Metab 2012; 97: 2670–81.

1μg=40IU, so 10μg=400IU, 20μg=800IU, and 1250μg=5000IU.

Personally, I'm not sure I agree. The mechanistic theory and correlations do not predict that 400IU is ideal, it doesn't seem enough to get blood serum levels of 25(OH)D to what seems optimal, and I don't even read Rejnmark the same way: look at the Figure 3 forest plot. To me, this looks like after correcting for Smith's use of D2 rather than D3 (D2 usually performs worse), that there are too few studies using higher doses to make any kind of claim (Table 1; almost all the daily studies use <=20μg), and the studies which we do have tend to point to higher being better within this restricted range of dosages.

That said, I cannot prove that 5k IU is equally or more effective, so if anyone is feeling risk-averse or dubious on that score, they should stick with 800IU doses.