Question

I’m trying to INNER JOIN a single row from a table. The query returns rows until I add the LIMIT 1 line of code and then it returns nothing and it doesn’t make any sense to me. Here is my code.

SELECT DISTINCT pd.pid, fv.flow_bmi, fv.flow_date
FROM patient_data pd
INNER JOIN ( ... )
INNER JOIN (
  SELECT pid, flow_bmi, flow_date
  FROM flow_vitals
  ORDER BY flow_date DESC
) fv ON fv.pid = pd.pid

#Results
-------------------------------
 pid  | flow_bmi  | flow_date
-------------------------------
 6073 | 31.90     | 2011-11-06
 6073 | 33.10     | 2011-11-05 <-- I don't want this row returned
 6072 | 32.10     | 2011-08-01

And now when I add LIMIT 1

SELECT DISTINCT pd.pid, fv.flow_bmi, fv.flow_date
FROM patient_data pd
INNER JOIN ( ... )
INNER JOIN (
  SELECT pid, flow_bmi, flow_date
  FROM flow_vitals
  ORDER BY flow_date DESC
  LIMIT 1
) fv ON fv.pid = pd.pid

#Expected Outcome
-------------------------------
 pid  | flow_bmi  | flow_date
-------------------------------
 6073 | 31.90     | 2011-11-06
 6072 | 32.10     | 2011-08-01

#Actual Outcome
-------------------------------
 pid  | flow_bmi  | flow_date
-------------------------------
 No rows returned.

I don’t know why LIMIT breaks this, and what I really want is an INNER JOIN that returns only one row based on the latest date. It seems like there are a ton of questions on this same topic but they end up using MAX() or LIMIT which both have not worked for me.

Answer 1

You actually want one row per group, not one row total from the flow_vitals table. Putting a LIMIT 1 there returns just the top row from the entire table.

I assume the row you want is the one with the maximum flow_date for that pid. Try:

SELECT DISTINCT pd.pid, fv.flow_bmi, flow_date
FROM patient_data pd
INNER JOIN ( ... )
INNER JOIN (
  SELECT pid, flow_bmi, flow_date
  FROM flow_vitals
) fv ON fv.pid = pd.pid
WHERE flow_date = (SELECT MAX(ffvv.flow_date) FROM flow_vitals ffvv WHERE ffvv.pid = fv.pid)

Or, as a join:

SELECT DISTINCT pd.pid, fv.flow_bmi, flow_date
FROM patient_data pd
INNER JOIN ( ... )
INNER JOIN (
  SELECT pid, flow_bmi, flow_date
  FROM flow_vitals
) fv ON fv.pid = pd.pid
INNER JOIN (
  SELECT pid, MAX(flow_date) AS maxDate
  FROM flow_vitals
  GROUP BY pid
) fvMax ON fv.pid = fvMax.pid AND fv.flow_date = fvMax.maxDate

I also wonder if DISTINCT is really necessary here.