使用子查询按特定列分组输出相同的错误结果答案

【问题标题】：Using sub query to group by specific column is outputting the same erroneous result使用子查询按特定列分组输出相同的错误结果
【发布时间】：2017-07-31 05:50:10
【问题描述】：

我在 SO 上询问了this question few days ago，但没有有效的答案。

问题在于以下查询：

SELECT t1.patient_id,
CONVERT(aes_decrypt(t4.patient_name_en, :encKey) USING utf8mb4) as patient_name_en,
min(t3.date_of_visit) as date_of_visit, 
t2.diagnosis_name,
max(ifnull(t5.date_of_assessment, 'N/A')) as date_of_assessment,
ifnull(t5.assessment_result, 0) as assessment_result 
FROM consultation t1
LEFT JOIN diagnosis t2 ON t1.diagnosis_id = t2.diagnosis_id
LEFT JOIN visit t3 ON t3.visit_id = t1.visit_id
LEFT JOIN patient t4 ON t4.patient_id = t3.patient_id
LEFT JOIN diabetes_assessment t5 ON t5.patient_id = t4.patient_id
WHERE t2.diagnosis_name LIKE :diagName AND t1.clinic_id = :cid
AND t3.visit_status=:visit_status
GROUP BY t1.patient_id, t5.date_of_assessment, t4.patient_name_en, t3.date_of_visit, t2.diagnosis_name, t5.assessment_result
ORDER BY t5.date_of_assessment DESC

给出以下结果：

但我真正想要的是仅按 patient_id 分组，这样我只能为每个早期诊断为糖尿病的患者获得一行。

我搜索了很多，发现我可以在这个查询中使用子查询。

所以我想出了以下子查询：

SELECT t1.patient_id
FROM consultation t1
LEFT JOIN diagnosis t2 ON t1.diagnosis_id = t2.diagnosis_id
LEFT JOIN visit t3 ON t3.visit_id = t1.visit_id
LEFT JOIN patient t4 ON t4.patient_id = t3.patient_id
LEFT JOIN diabetes_assessment t5 ON t5.patient_id = t4.patient_id
WHERE t2.diagnosis_name LIKE '%Diabetes%' AND t1.clinic_id = '361'
AND t3.visit_status="Active"
GROUP BY t1.patient_id

结果按patient_id分组，我得到了真正想要的2个id，而不是更多。即使是ID为0361的患者，每次去诊所就诊时都被诊断出患有多种糖尿病，但它的id只显示一次。

现在我想把这个子查询添加到最初的查询中：

SELECT t1.patient_id,
CONVERT(aes_decrypt(t4.patient_name_en, 'key1') USING utf8mb4) as patient_name_en,
min(t3.date_of_visit) as date_of_visit, 
t2.diagnosis_name,
max(ifnull(t5.date_of_assessment, 'N/A')) as date_of_assessment,
ifnull(t5.assessment_result, 0) as assessment_result 
FROM consultation t1
LEFT JOIN diagnosis t2 ON t1.diagnosis_id = t2.diagnosis_id
LEFT JOIN visit t3 ON t3.visit_id = t1.visit_id
LEFT JOIN patient t4 ON t4.patient_id = t3.patient_id
LEFT JOIN diabetes_assessment t5 ON t5.patient_id = t4.patient_id
WHERE t1.patient_id IN 
(SELECT t1.patient_id
FROM consultation t1
LEFT JOIN diagnosis t2 ON t1.diagnosis_id = t2.diagnosis_id
LEFT JOIN visit t3 ON t3.visit_id = t1.visit_id
LEFT JOIN patient t4 ON t4.patient_id = t3.patient_id
LEFT JOIN diabetes_assessment t5 ON t5.patient_id = t4.patient_id
WHERE t2.diagnosis_name LIKE '%Diabetes%' AND t1.clinic_id = '361'
AND t3.visit_status="Active"
GROUP BY t1.patient_id) AND
t2.diagnosis_name LIKE '%Diabetes%' AND t1.clinic_id = '361'
AND t3.visit_status="Active"
GROUP BY t1.patient_id, t5.date_of_assessment, t4.patient_name_en, t3.date_of_visit, t2.diagnosis_name, t5.assessment_result
ORDER BY t5.date_of_assessment DESC

但我得到了与上图相同的结果，其中每个示例 patient_id=0361 显示在 4 行中，但我希望显示一次，以及他第一次被诊断出患有糖尿病的日期。

【问题讨论】：

查询看起来一团糟，而且没有看到数据，我并不惊讶您之前没有得到任何答案。也许如果您可以将您的问题归结为更简单的问题，也许是只涉及 2-3 个表的连接，那么在这里获得帮助可能会更容易。
如果您听取了 cmets 提供的宝贵建议，或许可以
伙计们，我不知道如何创建 sql fiddles。另外，我有 6 张桌子。制作小提琴需要一个小时吗？

标签： mysql sql subquery

【解决方案1】：

SELECT t.patient_id,
CONVERT(aes_decrypt(t4.patient_name_en, 'key1') USING utf8mb4) as patient_name_en,
mindate as date_of_visit, 
tt2.diagnosis_name,
max(ifnull(tt5.date_of_assessment, 'N/A')) as date_of_assessment,
ifnull(tt5.assessment_result, 0) as assessment_result 
FROM consultation t
LEFT JOIN visit tt3 ON tt3.visit_id = t.visit_id
LEFT JOIN diagnosis tt2 ON t1.diagnosis_id = tt2.diagnosis_id
INNER JOIN
(
    SELECT t1.patient_id,min(t3.date_of_visit) mindate
    FROM consultation t1
    LEFT JOIN diagnosis t2 ON t1.diagnosis_id = t2.diagnosis_id
    LEFT JOIN visit t3 ON t3.visit_id = t1.visit_id 
    WHERE t2.diagnosis_name LIKE '%Diabetes%' AND t1.clinic_id = '361'
    AND t3.visit_status="Active"
    GROUP BY t1.patient_id
)INNERTABLE ON t.patient_id=INNERTABLE.patient_id and INNERTABLE.mindate
LEFT JOIN patient t4 ON t4.patient_id = tt3.patient_id
LEFT JOIN diabetes_assessment t5 ON t5.patient_id = t4.patient_id

试试上面的查询。

【讨论】：

我会尝试并回复您。

【解决方案2】：

你的最终目标是什么？

只获取 Patient_id、date_of_assessment 吗？那么你只需要“分组”1列。喜欢：

select patient_id, min(date_of_assessment) from .... where ....  group by patient_id

如果您需要在同一列上列出诊断，则需要使用特定的 mysql 事物，例如：

select patient_id, min(date_of_assessment), GROUP_CONCAT(diagnosis SEPARATOR ' ') from .... where ... group by patient_id

问题是您需要为数据库提供一种方法来计算您未分组的列的值。可以是 sum, count, min, max, ....

【讨论】：