Medicare doctor charge database query

doctors.sql

SELECT medicare.*, 
       peerAverage.average       AS charge_average, 
       ( medicare.average_submitted_chrg_amt - peerAverage.average ) / 
       peerAverage.average * 100 AS difference_from_average, 
       totalAverage.total_difference_average 
FROM   medicare 
       JOIN (SELECT Avg(average_submitted_chrg_amt) AS average, 
                    hcpcs_code, 
                    place_of_service 
             FROM   medicare 
             WHERE  npi LIKE '%' 
                    AND hcpcs_code LIKE medicare.hcpcs_code 
                    AND nppes_provider_city LIKE 'Latrobe' 
                    AND nppes_provider_state LIKE 'PA' 
             GROUP  BY place_of_service, 
                       hcpcs_code) AS peerAverage 
         ON medicare.hcpcs_code = peerAverage.hcpcs_code 
            AND medicare.place_of_service = peerAverage.place_of_service 
       /* This join basically repeats everything and to get total percentage average by doctor */
       JOIN (SELECT npi, 
                    Avg(( medicare.average_submitted_chrg_amt - 
                          peerAverage.average ) / 
                            peerAverage.average * 100) AS 
                    total_difference_average 
             FROM   medicare 
                    JOIN (SELECT Avg(average_submitted_chrg_amt) AS average, 
                                 hcpcs_code, 
                                 place_of_service 
                          FROM   medicare 
                          WHERE  npi LIKE '%' 
                                 AND hcpcs_code LIKE medicare.hcpcs_code 
                                 AND nppes_provider_city LIKE 'Pittsburgh' 
                                 AND nppes_provider_state LIKE 'PA' 
                          GROUP  BY place_of_service, 
                                    hcpcs_code) AS peerAverage 
                      ON medicare.hcpcs_code = peerAverage.hcpcs_code 
                         AND medicare.place_of_service = 
                             peerAverage.place_of_service 
             WHERE  medicare.npi LIKE '%' 
                    AND nppes_provider_last_org_name LIKE '%' 
                    AND nppes_provider_first_name LIKE '%' 
                    AND nppes_provider_city LIKE 'Pittsburgh' 
                    AND nppes_provider_state LIKE 'PA' 
                    AND medicare.hcpcs_code LIKE '%' 
             GROUP  BY npi) AS totalAverage 
         ON medicare.npi = totalAverage.npi 
GROUP  BY npi, 
          place_of_service, 
          hcpcs_code 
ORDER  BY hcpcs_code ASC, 
          difference_from_average DESC 

@arm5077 Try throwing an index on the nppes_provider_state column of the averages table as well. It's the only thing I can think of that might be slowing it down. How many rows ended up in the averages table?

I'm more familiar with postgres, so I might pull this down into a postgres db and give it a try. I also have access to some Really Honking Powerful Postgres servers, so if i get it working locally, I might give the query a try there as well.

@seejee Will try the index. The averages table has 2 million rows ... more than I expected, but logical, I guess -- you're dealing with 6,000 procedures across 50 states and 18,000 cities.