Dear Experts,
if the average number of medical records per week 46 records , and I want to select a weekly random sample , may I know how to identify the weekly sample size?
Many thanks.
Omar.

Dear Experts,
if the average number of medical records per week 46 records , and I want to select a weekly random sample , may I know how to identify the weekly sample size?
Many thanks.
Omar.

By the way - if you are counting defects per record, it will be a u chart. If you are calculating the rate of records with at least one defect, then that will be a p chart.

It depends. Do you have a standard that you must achieve? For example, if you want to be 95% sure that no more than 5% of the records are defective, you need to sample 59 records and get no defects. Now, this is made more interesting by the fact that this is a time series of events - 46 records per week. So perhaps you might choose to be sure you sampled 59 per month so that you will know after no longer than a month that you have exceeded the 95/5 threshold.

You see, in general the question of how often and how much to sample is driven by balancing two costs

1. How much does it cost to do a sample?

2. How long are you willing to go with an increased level of error without detecting it? What are the losses per bad record?

Personally, I offhand try for 25 per month as a sample. It gives me enough for a useful monthly p or u chart with reasonable control limits to detect trends.