We have over 10 years of data on a number of hospitals that we regularly transport patients to. This past February, a new hospital (1830) opened that has substantially altered our transport patterns, especially taking away from the previous primary destination (1247).

Destination   Dec     Jan     Feb     Mar     Apr
1247          117     140     131      65      77
1017           75      69      54      45      31
1818           56      58      43      46      39
1830            -       -      30     125      86     

I want to evaluate the overall impact (improved in-service times, reduced mileage/vehicle costs, etc...) of the new facility. What is the appropriate manner to evaluate the new facility? Should I compare equal periods prior to and after it opened? Or, can I evaluate the prior year and the period since it opened? Any help is appreciated.

EDIT: I'm not sure that I have too much seasonal/cyclical variance in my long-term data. Here's a chart that has the previous 5 year history (each point is a 2-week period). I've highlighted total patients transported (with 26 period moving average), and the destination hospital in question.

Patient Transport History

  • $\begingroup$ I should have mentioned that I use 28-day "months" for reporting to avoid issues related to number of days in the reporting period. $\endgroup$
    – dav
    Commented Jun 5, 2012 at 12:15

1 Answer 1


As I understand you have some strong seasonal prevalence in your data. In this case I'd rather do the following things:

  1. It is always better to compare the period you want to analyse (since February till now) to the same period last year to get rid of seasonal effect.
  2. It is better to compare not the absolute values, but the percents (just divide everything by the total amount). This will get rid of year-to-year difference (for example average increase of costs).

Hope it will help.

  • $\begingroup$ Thanks for the feedback-I've added some longer term numbers to explore the seasonality angle. I was concerned about using a prior year's period because of the impact of long-term trends, but you covered that in #2. $\endgroup$
    – dav
    Commented Jun 5, 2012 at 17:04
  • $\begingroup$ This data is surprisingly stable! I was almost sure that there should be seasonal effect. In this case I have seen reports that providing both numbers: "year-to-year period" comparison, and "period-to-the previous one". If both numbers (relative numbers) show the improvement - you have nothing to explain any more (you cover both short- and long-term improvement). $\endgroup$ Commented Jun 8, 2012 at 15:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.