Just curious, why no pipes to uptime? You can capture the output from uptime, or the 'w |head -1' command to get the load average very quickly, and seemingly reliably on unix systems (which I assume you are on due to the reference to /proc)
If you have those limitations, what else can you use or not use?