I've never seen such a thing with PRINTING! Reading data, ok, that's sometimes a problem. Are you sure, that there are no strange (invisible) characters in the field instead of a blank? That's what I assume! Perhaps you could try to print that obs and variable with a $hex format, or browse the raw-file with a hex-browser or editor to see whats in. If you identify something strange, you should use COMPRESS with all the wrong chars to remove them.

On Thu, 23 Feb 2006 07:30:52 -0800, A Pal <nbertest@YAHOO.COM> wrote:

>Hi All, > > MISSOVER and TRUNCOVER are giving the same result. > There is no "INPUT went to a new line..." in the log. The data is reading successfully so that > > proc print; > var contract_id -- step_ther_yn ; ** everything but the drug_name ; > > looks correct and normal: > > 8 H0104 00001356 9 00002950501 1 N N N > 9 H0104 00001356 9 00002721701 1 N N N > 10 H0104 00001356 9 00013105602 1 N N N > 11 H0104 00001356 9 00006494300 1 N N N > > but printing drug_name only seems to produce an overwrite like below: > > proc print; > var drug_name; > > 8 CEFAZOLIN SODIUM > 9 HEPARIN SODIUM > > PNEUMOVAX 23 > > ( Note, no Obs for 10 and 11 ). > > Thanks so much for help! > > Jyan > > Gerhard Hellriegel <ghellrieg@T-ONLINE.DE> wrote: did you try the infile - option MISSOVER or TRUNCOVER? > >... >infile xxx dlm="|" missover; >... > >that should prevent the input statement to read a new line, if there is no >data left. >Did you see any "INPUT went to a new line..." ? > > > > >On Thu, 23 Feb 2006 06:40:19 -0800, A Pal wrote: > >>Hi All, >> >> Thank you all so much for the helpful responses! Using the : in the input >> statement for the NDC code variable cause the tier variable to be read >> correctly. >> >> I also have a question about how to read the drug name variable. It is >> the last variable in the medicare public use formulary data file. It ends >with >> a line feed \n character, and if there is no drug name, then the value is >a single >> space. I am reading it with drug_name : $40. in the input statement. The >> problem I find is that raw data like this: >> >> H0104|00001356|9|00002950501|1|N|N|N|CEFAZOLIN SODIUM >> H0104|00001356|9|00002721701|1|N|N|N|HEPARIN SODIUM >> H0104|00001356|9|00013105602|1|N|N|N| >> H0104|00001356|9|00006494300|1|N|N|N|PNEUMOVAX 23 >> >> is displayed by PROC PRINT like this: >> >> 8 H0104 00001356 9 00002950501 1 N N N CEFAZOLIN SODIUM >> H0104 00001356 9 00002721701 1 N N N HEPARIN SODIUM >> 6 9 00013105602 1 N N N >> 0104 00001356 9 00006494300 1 N N N PNEUMOVAX 23 >> >> ( The leading '8' is the Obs ). The variables are all seeming to be read >> correctly. However, the drug_name variable seems to be overwriting the >> subsequent line. Another analyst, not sure who but probably not experienced >> with SAS, will be running this program each month when the data is updated. >> So I would like that person to have no surprise when they run this program. >> ( I tried using a length drug_name $40 ; prior to the infile statement. I >can verify via proc contents that SAS is implementing it, however,it doesn't >seem to be helping. ). Have others encountered a problem like this and >solved it? >> >> Thanks, >> >> Jyan >> >>Jiann-Shiun Huang wrote: Jyan: >> >> I believe the problem is in the input statement: >> >>input contract_id $ formulary_id $ formulary_version ndc $11. cheesy >>tier >> quantity_limit_yn $ prior_auth_yn $ step_ther_yn $ drug_name : $40. ; >> >>You use list input to get your variable values and hence you should >>have a colon between "ndc" and "$11.". With the change above and take >>out cheesy from the input statement, the input statement is: >> >>input contract_id $ formulary_id $ formulary_version ndc:$11. tier >> quantity_limit_yn $ prior_auth_yn $ step_ther_yn $ drug_name : $40. ; >> >> Without ":" between "ndc" and "$11.", the column pointer stops at "|" >>for the next input. That is why your "cheesy" variable get a missing >>value since a nonnumeric value is tried to be assigned to a numeric >>variable "cheesy." >> >> >>J S Huang >>1-515-557-3987 >>fax 1-515-557-2422 >> >>>>> A Pal 2/22/2006 4:04:19 PM >>> >>Hi All, >> >> I'm trying to infile a pipe delimited public use medicare formulary >>file. >> The record layout has an 11-digit NDC code followed by a one-digit >> drug tier code. I found that reading it in order using the code >>below >> made the 'tier' drug get set to missing and all subsequent variables >> were shifted over. The last variable is drug name and it can be 40 >> characters long, with spaces, and if there is no drug name, the >> field is filled with a single space. This datafile has over 10 >>million >> records and it is replaced every month, so it would be great to have >> a working program that doesn't resort to the cheesy trick of >>including >> the 'cheesy' variable as I've done below and to make the drug name >> variable print correctly ( obs gets overwritten in 10 and 11 ). >> Many thanks for help! I've been playing with reading the data >> in a number of different ways, and this is the best I've been able to >>come up with. >> >> Thanks, >> >> Jyan >> >> RAW DATA: >> >> H0104|00001356|9|00002950501|1|N|N|N|CEFAZOLIN SODIUM >> H0104|00001356|9|00002721701|1|N|N|N|HEPARIN SODIUM >> H0104|00001356|9|00013105602|1|N|N|N| >> H0104|00001356|9|00006494300|1|N|N|N|PNEUMOVAX 23 >> >> CODE: >> >> data formulary ; >> infile "mydata.txt" dsd delimiter='|' ; >> input contract_id $ formulary_id $ formulary_version ndc $11. cheesy >>tier >> quantity_limit_yn $ prior_auth_yn $ step_ther_yn $ drug_name : $40. ; >> >> RESULTS: >> >> 8 H0104 00001356 9 00002950501 . 1 N N N >> 9 H0104 00001356 9 00002721701 . 1 N N N >> 10 H0104 00001356 9 00013105602 . 1 N N N >> 11 H0104 00001356 9 00006494300 . 1 N N N >> >> Obs drug_name >> >> 8 CEFAZOLIN SODIUM >> 9 HEPARIN SODIUM >> >> PNEUMOVAX 23 >> >> >> >> >> >> >> >> >> >> >>--------------------------------- >>Brings words and photos together (easily) with >> PhotoMail - it's free and works with Yahoo! Mail. >> >> >> >>--------------------------------- >> Yahoo! Mail >> Use Photomail to share photos without annoying attachments. > > > >--------------------------------- >Brings words and photos together (easily) with > PhotoMail - it's free and works with Yahoo! Mail.