Limit for days_supply, refills and quantity

Olga_Osintseva · July 2, 2019, 12:59pm

Hi!
In Achilles Heel Report i find warning about limit of days_supply, refills and quantity.

WARNING: Distribution of days_supply by drug_concept_id; max value should not be > 180
WARNING: Distribution of refills by drug_concept_id; max value should not be > 10
WARNING: Distribution of quantity by drug_concept_id; max value should not be > 600

Does anybody know anything about these limitations? How are they determined?

Thank you in advance

Chris_Knoll · July 2, 2019, 1:32pm

The rules are defined here (bookmarked to the days supply rule definition):

github.com

OHDSI/Achilles/blob/d5a0ff786ae219d8773d9d45332ca51a682cd017/inst/csv/heel/heel_rules_all.csv#L23


12,Dx is not a SNOMED code,parallel,heel_results,error,CDM conformance,"concept from the wrong vocabulary; Condition Occurrence, Condition Era (SNOMED)",,,,400;1000
13,Drug is not RxNorm concept,parallel,heel_results,error,CDM conformance,"concept from the wrong vocabulary; Drug Exposure, Drug Era (RxNorm)",,,,700;900
14,"Procedure is not CPT, ICD9Proc or HCPCS",parallel,heel_results,error,CDM conformance,procedure - 4 CPT4/5 HCPCS/3 ICD9P,,,,600
17,incorrect terminology,parallel,heel_results,error,CDM conformance,revenue code - 43 revenue code,,,,1610
18,year of birth is in the future,parallel,heel_results,error,DQ,year of birth should not be in the future ,,plausibility,,3
19,year of birth is prior 1800,parallel,heel_results,warning,DQ, year of birth < 1800,<1800,plausibility,,3
20,age below 0,parallel,heel_results,error,DQ,age < 0,<0,plausibility,,101
21,age too high,parallel,heel_results,error,DQ,age > 150,>150,plausibility,,101
22,monthly trend,parallel,heel_results,warning,DQ,monthly change > 100%,,fidelity,,
23,monthly trend,parallel,heel_results,warning,DQ,monthly change > 100% at concept level,,fidelity,,
24,too high days_supply,parallel,heel_results,warning,DQ,days_supply > 180,,plausibility,,
25,too high number of refils,parallel,heel_results,warning,DQ,refills > 10,>10,plausibility,,716
26,implausible quantity for drug,parallel,heel_results,warning,DQ,quantity > 600,>600,plausibility,,717
27,more than 1 percent of unmapped rows (concept_0 rows),serial,both,warning,DQ,for multiple analyses (4xx;6xx;7xx;8xx;18xx),>1,completeness,,UnmappedData:byDomain:Percentage
28,percentage of non-numerical measurement records exceeds general population threshold,serial,both,warning,DQ,"typically, measurement data contans a significant proportion of rows with numerical result. This rule looks at rows in MEASUREMENT and alerts the user if a large proportion of rows lack any numerical result",>=80,completeness,GeneralPopulationOnly,Meas:NoNumValue:Percentage
29,infant diagnosis at senior age of over 50yo,serial,heel_results,error,DQ,mecconium condition 195075; This rule is example of a terminology depended data quality tool,,plausibility,,404
31,ratio of providers to total patients,serial,both,notification,DQ,This rules fires if data indicate a high number of patients and only a few providers exist. ,,plausibility,,Provider:PatientProviderRatio
32,Percentage of patients with no visits exceeds threshold,serial,heel_results,notification,DQ,checks if the percentage of patients with no visits exceeds threshold,>5,plausibility,,ach_2003:Percentage
33,[GeneralPopulationOnly] Not all deciles represented at first observation,serial,heel_results,notification,DQ,"in a general population, a database would observe first visit across all age groups. We at least expect deciles 0 to 8. Rule looks at the count of deciles.",<9,completeness,GeneralPopulationOnly,AgeAtFirstObsByDecile:DecileCnt
34,Count of unmapped source values in a domain exceeds threshold,serial,heel_results,notification,DQ,"looks at values that are mapped to concept0 and their source values by table, rule 6 is related to this rule but it does not look at the size of the problem (only if unmapped data are present or not present)",,completeness,,UnmappedDataByDomain:SourceValueCnt
35,Count of measurement_ids with more than 5 distinct units  exceeds threshold,serial,heel_results,notification,DQ,"Idealy, each measurement would use only one unit. For example, kg for weight. This rule notifies the user if database has measurements that have 5 or more units. This rule has technically thresholds. ",>=5;>=10,fidelity,,1807

The sourcecode to that rule is here:

github.com

OHDSI/Achilles/blob/d5a0ff786ae219d8773d9d45332ca51a682cd017/inst/sql/sql_server/heels/parallel/heel_results/rule_24.sql

--ruleid 24 WARNING: days_supply > 180 

--HINT DISTRIBUTE_ON_KEY(analysis_id)
select
	analysis_id,
	ACHILLES_HEEL_warning,
	rule_id,
	record_count
	into @scratchDatabaseSchema@schemaDelim@tempHeelPrefix_@heelName
from
(
  SELECT DISTINCT ord1.analysis_id,
    CAST(CONCAT('WARNING: ', cast(ord1.analysis_id as VARCHAR(10)), '-', oa1.analysis_name, ' (count = ', cast(COUNT_BIG(ord1.max_value) as VARCHAR(19)), '); max value should not be > 180') AS VARCHAR(255)) AS ACHILLES_HEEL_warning,
    24 as rule_id,
    COUNT_BIG(ord1.max_value) as record_count
  FROM @resultsDatabaseSchema.ACHILLES_results_dist ord1
  INNER JOIN @resultsDatabaseSchema.ACHILLES_analysis oa1
  	ON ord1.analysis_id = oa1.analysis_id
  WHERE ord1.analysis_id IN (715)
  	AND ord1.max_value > 180

This file has been truncated. show original

In general, the way the heel reports work is that they look to the summary statistics (the maximum value of days supply, in this example) and will output a warning if the value is suspicious (in this case, a DRUG_EXPOSURE Record with a days supply of 180 is suspicious).

Olga_Osintseva · July 2, 2019, 1:55pm

Thank you!

Alexandra_Orlova · July 17, 2019, 2:13pm

Do you have any idea why Achilles limits days_supply / refills / quantity using exactly those numbers?
For example, Achilles states that quantity shouldn’t be more than 600. What does 600 mean? Tablets, milliliters or what? Each drug has its own units, not sure that it is correct to use such limitation. What do you think?
Same about refills and days supply. Do you know why those particular numbers are used? maybe, they are from some medical documentation?

julie_kohler · July 17, 2019, 2:57pm

I think the relevant unit is days. So, patients don’t get more than about 2 years worth of drugs at a time (which seems pretty generous, in my experience).

Alexandra_Orlova · July 17, 2019, 3:13pm

the days_supply field captures info about a duration of taking a drug and this field has Achilles limit equaled 180 days, so according to Achilles patient cannot get more than half of a year worth of drugs at a time. And I’m wondering why 180 days limit is using? Same for refills.

the quantity field captures info about the quantity of drug (10 tablets, 100 mg, etc.) and this field has Achilles limit equaled 600. I cannot understand this limit at all. As I mentioned above each drug has its own units, but this limit is applied to all drugs. Is it correct?

Christian_Reich · July 17, 2019, 3:29pm

Because usually patients don’t get prescribed years worth of prescription drugs. They have to come back in to renew the prescription. There are exceptions, though (folks moving to some developing country or so), but that is rare.

True, but only half true. If the Drug Concept is an ingredient or Drug Component then the quantity is whatever unit the ingredient is assigned to. You can find that in DRUG_STRENGTH. We are revising the CDM and will explain that painstakingly. Right now it is all over the place.

Christian_Reich · July 17, 2019, 3:29pm

So:

Days_supply - days
Refills - number without a unit
Quantity - sales unit if Clinical Drug or less, measurement unit if Ingredient or Drug Component.

Alexandra_Orlova · July 18, 2019, 8:11am

Thank you for clarifying unit logic.
So I believe, it’s not a good decision to apply the same limitation (600) for all types of drugs because quantity depends on duration (days_supply) and units.
Example 1:
patient received the prescription: “take Aspirin 325 MG Oral Tablet [Aspir-trin] one tablet 2 times a day for 20 days”:
days_supply = 20
quantity = 1* 2 * 20 = 40 tablets

Example 2:
patient is administered the drug: “Treprostinil Injectable Solution [Remodulin] 15.5mg one time a day for 13 days”:
days_supply = 13
quantity = 15.5mg * 1 * 13 = 201.5 mg

Achilles applies the same limitation = 600 for both examples. It means that quantity1 (40 tablets) should be <= 600 and quantity2 ( 201.5 mg) should be <= 600. It doesn’t look correct, does it?
I suppose limitation should be based on days supply and unit of drug. I think it is possible to have a drug with quantity > 600 if the duration of taking the drug is long-lasting and units are mg. Probably, it’s better to have a limitation of daily dose? Am I missing something?