I have a SAS output data set of drugs prescribed for a Medicaid population.   I have two issues: 1) setting up a if-then-else section to search for specific hiv/aids drugs by brand and/or generic. The following is the coding. The brand name is mentioned first and the generic is second. Another wrinkle is that there is one brand but (many times) several generics (or more like generic cocktail mix). (There are many lines so I am going to shorten the example. )   Data hivaids.hivaids_drugs (keep = DSC_NDC label);     set hal_data.mcaid_drug_dn;     If dsc_ndc eq "Emtriva" or "emtricitabine" then label eq "Yes";    Else If dsc_ndc eq "Delstrigo" or "doravirine" and "lamivudine" and    "tenofovir disoproxil fumarate" then label eq "Yes"; : : : Else If dsc_ndc eq "Kaletra" or "lopinavir" and "ritonavir" then label eq "Yes";   So the error I run into is INVALID numeric data, << drug name >> at line xxxx column xx   That error repeats many times.  All the drug names on the right hand side of the  OR are the generics and there can be from two to five. I separated each by AND. I received output which was basically just the column DSC_NDC in the SAS output set.  What I really want is to match the list of hiv/aids drugs against the SAS output set and then only keep those where label = "Yes".   That is the first problem.   Secondly in the SAS output set there may be several lines for the same drug depending upon how the drug is administered. For example,        Oncovin             Vial        Oncovin              Syringe       Oncovin               Syringe   I just want to identify the name once - but I get every line for each drug by mode of delivery as listed in the SAS dataset.   So the second problem is just having the name of any drug match listed only once.   Thank you for you assistance on this. Walt Lierman ... View more

I have nine diagnosis code attributes (columns) labeled Diag1, Diag2,  ..... , Diag9   I have several icd-10 dx_codes that I need to check across 3 years of Medicaid recipient claims   Here is the c ode starting after the FROM clause   WHERE Month Between '201701' And '201912'     And ClaimType Not In ('D')     And  Diag1 - Diag9 In ('B20', 'B9735', ........., '09873') ORDER By Recipno, Month; quit;   The problem occurs at Diag1- Diag9  with the error reading: Expression using negation (-) requires numeric types   I have tried several variants of the Diag1-DIag9 construct but none have solved the issue. I appreciate your help again.   Walt Lierman ... View more

I have written the following short query   proc sql noprint; CREATE table aqc_puds.data_count_claim AS SELECT               claimtypedesc,               claimtype,               Count(claim) FROM aqc_puds.base_yr_pudset GROUP BY claimtypedesc; quit;   Desired output data (hypothetical numbers): Claimtypedesc           count dental                         1534236 inpatient                   22342522   professional             44999900             ....... and so on ...... there should be about 10 rows - one for every claimtype - with claims counted over each claimtype   actual output data Claimtypedesc              count dental                            15151016 dental                            15151016    .                                       .    .                                       .    .                                       .   and so on for each dental claim   So the dilemma is to obtain the first hypothetical table rather than the erroneous actual table.   This is likely a pretty straightforward solution.  However, I haven't been able to crack the code and so your expertise is very appreciated.   Walt Lierman    ... View more

  I have a dataset with 3.5 million obs and only three variables: recipient_ID, claim, month (in form like '201704' for example)   Data Stg_Out4.PMCA_Claims_table_3 (drop = tcn month); Set Stg_Out4.Claims_table_2_2019; By recipno tcn; retain Recipno Month Claim_201607 Claim_201608 Claim_201609 Claim_201610 Claim_201611 Claim_201612 Claim_201701 Claim_201702 Claim_201703 Claim_201704 Claim_201705 Claim_201706; Array clmmon (12) Claim_201607 Claim_201608 Claim_201609 Claim_201610 Claim_201611 Claim_201611 Claim_201612 Claim_201701 Claim_201702 Claim_201703 Claim_201704 Claim_201705 Claim_201706; Run;   I have executed proc transpose to put the claim_month as a header line.  I created an array structure (see code above).    I have created flags ('1') for the months that had a claim by each recipient in this step:   /** Create flags **/ Data Stg_Out4.Claims_flags_2019; Set Stg_Out4.PMCA_Claims_table_3; by Recipno; If Claim_201607 ne '.' then Claim_201607 = '1'; else If Claim_201608 ne '.' then Claim_201608 = '1'; else If Claim_201609 ne '.' then Claim_201609 = '1'; else If Claim_201610 ne '.' then Claim_201610 = '1'; else If Claim_201611 ne '.' then Claim_201611 = '1'; else If Claim_201612 ne '.' then Claim_201612 = '1'; else If Claim_201701 ne '.' then Claim_201701 = '1'; else If Claim_201702 ne '.' then Claim_201702 = '1'; else If Claim_201703 ne '.' then Claim_201703 = '1'; else If Claim_201704 ne '.' then Claim_201704 = '1'; else If Claim_201705 ne '.' then Claim_201705 = '1'; else If Claim_201706 ne '.' then Claim_201706 = '1'; run; So my question (likely very simple) when I open the flagged data set. The header line is not in calendar order instead it is:   Claim_201607  Claim_201608 Claim_201609  Claim_201612 CLAIM_201704 Claim_201705 Claim_201706 Claim201610   Claim_201611 Claim_201701 Claim_201702  Claim_201703    instead of the correct order.   Can you please show the corrected approach to get the header labels in proper order.  Is the fix universal in that it will work for other label types?   Thank you for your assistance.   Wlierman ... View more

I am trying to execute a query to search for keywords from ICD10 codes and receive the error written in the Subject line. My query follows:   Proc sql noprint; Create table NCI_Out.Conditions_Flags As Select Distinct RecipientID, ClaimType As Condition Label 'Condition', Gender, Medicare_status, Month, CASE When Primary_Dx_Desc Like '*Depress%' Then Condition = 'Depression' Else ' ' End As BI12_01, Case When Primary_Dx_Desc Like '*Schizo%' Then Condition = 'Schizophrenia' Else ' ' End As BI12_02, Case When Primary_Dx_Desc Like '*BiPol%' Then Condition = 'BiPolar' Else ' ' End As BI12_03, Case When Primary_Dx_Desc Like '*Mood%' Then Condition = 'Mood_Disorder' Else ' ' End As BI12_04, Case When Primary_Dx_Desc Like '*Anxiety%' Then Condition = 'Anxiety_Disorder' Else ' ' End As BI12_05, Case When Primary_Dx_Desc Like '*Chemical%' Then Condition = 'Chemical_Dependence' Else ' ' End As BI12_06, Case When Primary_Dx_Desc Like '*Opiate%' Then Condition = 'Chemical_Dependence' Else ' ' End As BI12_06a, Case When Primary_Dx_Desc Like '*Opiod%' Then Condition = 'Chemical_Dependence' Else ' ' End As BI12_06b From NCI_Out.ICD10_Text_Check_ok Order by RecipientID, Month; Quit;Proc sql noprint;   I appreciate any help that is offered. ... View more

I have a dataset with just under 59K obs.  This data are Medicaid recipients. There are 18 Medicaid CCOs and one open-card designation (these recipients are most likely 'between' membership in a CCO) . The observations are divided between the CCOs on the basis of geographical location within the state. In earlier code a continuity of care measure was created.    I want to create some simple column charts that will group the CCOs together by continuity of care values. The index values range between 0 and 1.  As I envision it I would like the following                                                --------------------- Index_score across the top ---------------------------                                               1            2        3         4           5           6          7            8           9                              CCO by name        <down the left side>                  <the body of the table would contain the counts of the                                                            scores - that is, how many recipients in each CCO and the Open-card                                                            fall into index range 1 through 9 >   This is the code so far (using %Let statements to hold whatever index values I want to create a chart)   %Let Lb_0 = 0.000; %Let Lb1 = 0.0001; %Let Lb2 = 0.2499; %Let Lb_25 = 0.250; %Let Lb3 = 0.250001; %Let Lb4 = 0.4999; %Let Lb_50 = 0.500; %Let Lb5 = 0.50001; %Let Lb6 = 0.7499; %Let Lb_75 = 0.7500; %Let Lb7 = 0.750001; %Let Lb8 = 0.9999; %Let Lb_1 = 1.000;   Data StgCOCxx.CCO_COC_Range (keep = CCO COC_BB_Index indscr);     Set StgCOCxx.COC_Pmca_complex_Detail_Final; by CCO; *retain indscr CCO; do i = 1 to 9;   if COC_BB_Index = &Lb_0 then indscr = 1;   if &Lb1 <= COC_BB_Index <= &Lb2 then indscr=2;   if COC_BB_Index = &Lb_25 then indscr=3;   if &Lb3 <= COC_BB_Index <= &Lb4 then indscr=4;   If COC_BB_Index = &Lb_50 then indscr=5;   If &Lb5 <= COC_BB_Index <= &Lb6 then indscr=6;   If COC_BB_Index = &Lb_75 then indscr=7;   If &Lb7 <= COC_BB_Index <= &Lb8 then indscr=8;   If COC_BB_Index = &Lb_1 then indscr=9; end; run;   My abortive attempt on the next data step where I want to create the array   Data StgCOCxx.COC_array_index;   Set StgCOCxx.CCO_COC_Range;           array index[*]_numeric_;                   .......................   I started reading papers on Arrays and ended up restarting a couple of times.  I thoroughly have myself confused.   Your help will be greatly appreciated.   wlierman ... View more