Okay, I’m back from visiting the parental units and enjoying yet another show of how the Granddaughter Rules play out. My dad was always involved with us, but his involvement usually stopped at baths, balls, books, banks and trips to the airport. That was where the combination of his knowledge base and societal expectations of blue collar manliness met. This weekend I saw him (attempt to) braid a doll’s hair while we were waiting for a table at a restaurant and get a complete make-over. The Granddaughter rules.
The big health insurance news this weekend was the 10% error rate on the 834 files that the Feds are generating. Some of this seems to be systemic errors, which are extremely fixable. Most of the problems seem to be a combination of human errors and system design problems. Those are a bit tougher to fix but are also fixable:
The Los Angeles Times reports:
Insurers have seen various types of errors in the 834 transmissions, including garbled or incorrect information, duplicate forms and, in some cases, missing forms.
The Wall Street Journal has details on error types:
A source familiar with the project blamed 834 problems on the use of poor software “rules,” essentially a system of programming routines that govern how the marketplace application responds to each piece of data consumers enter in a HealthCare.gov form. For example, if a consumer accidentally puts an ampersand in the address box in their form, rather than receive an error, they will be allowed to continue filling out their application online. However, the ampersand will prevent the 834 file from being generated. “Software rules intended to prevent certain errors weren’t put into place,” the source said. “People are calling up health plans, saying they have enrolled, and [insurers] have no record of them,” the source said.
Data integrity between multiple interfacing systems with varying formating rules is a pain in the ass. I’m currently working on a different project that is attempting to marry a location to a claim through externally maintained data structures. Most of the initial work has been merely making sure that my company’s Address 1 maps and parses correctly to the external data structure’s Address Location 1 field.
I’m not too worried that 834s are erroring out. My company has an entire department dedicated to cleaning up 834s because they are almost never perfect for any firm of more than 3 people and a single contract. I’m worried that some of the common rules were either not written into the interfacing layer of the data hubs OR a pre-exisiting off the shelf 834 validation system was not integrated into the build. Weird things will happen but the commonly strange things should be accounted for and cleaned out before they impact members or prospective members.
OzarkHillbilly
Ahh yes, Granddaughter Rules, I am very familiar with them. She makes the rules, and if PawPaw knows what’s good for him, and I do, he does what he is told.
Oh wait a minute, you mean that the Granddaughter rules. Well, yeah. I mean like, Duuuhhhhhh, Same thing.
MomSense
I was just thinking the other day how much I miss my grandfather. He had the warmest smile in the whole world and we had some wonderful walks together.
Linda Featheringill
@MomSense:
Your grandfather:
What nice memories.
Actually, I have found that the missing doesn’t ever really, really go away. Like if you had a leg amputated, you might go on with life but you’d always miss that leg.
Hugs.
dpm (dread pirate mistermix)
In the WSJ article, it says “Insurers have no record of them” – is that what’s really happening? Or is it that their application is in a queue somewhere at the insurer waiting to be fixed by hand? It’s a bigger problem if the application hadn’t reached the insurer. If it’s there but hasn’t been entered yet because of a data issue, that’s probably less of an issue.
Cervantes
@MomSense: It’s good to remember.
And in turn to build memories for the ones we will leave behind.
Richard Mayhew
@dpm (dread pirate mistermix): The 834 is not generated as the data fails validation at healthcare.gov/HHS/Center for Medicare Services so the insurance company does not know about the member sign-up.
This is correctable (mainly through cross-walking identifiers and saying I’ve got SSN ZZZ-XX-YYYY, DOB 1/12/1955, M, saying he is at Mayhew Insurance PPO Preferred Supreme — Mayhew Insurance — do you have him this record? Yes
Okay, what about…..
El Caganer
So they didn’t have any software rules that would generate an error message for an obviously incorrect entry such as an ampersand. Jesus Christ. Who did they contract the coding out to, the Missouri Primate Foundation?
MomSense
@Linda Featheringill: @Cervantes:
Thanks. We have this license plate here in Maine that supports small farms and it shows a farmer holding hands with a little girl. I spent many a beautiful summer day walking hand in hand with my grandfather on his farm.
The longer I live the more I think it is the simple things that are the most important and memorable.
Another Holocene Human
@El Caganer: Not only that, plenty of websites will do a check to validate your address with the USPS. Does the USPS charge for that or something? (And whose fault is that… GOP?)
Not stripping escape characters is fucking coding 101. That’s kinda sorta sad.
But then, take all the people who fumbled their SSN and couldn’t fix it without trashing the app and starting over? That’s crazy.
burnspbesq
When I signed up with BS of CA via Covered CA on 11/8, the last screen said that I would get an enrollment package and payment instructions in the mail on or about 11/15.
When it didn’t show up by last week, I got concerned, so I called BS directly last week, and was told yes, they have me in the system. The rep even gave me a case number, and promised that the enrollment package would be mailed this week.
There are stresses all though the system because of the unprecedented volume of activity.
Nicole
My stepmom, who has five grandkids and three step grandkids, has a pillow on the chair that says “If I’d known grandkids were this much fun I would’ve had them first.” Always makes me laugh.
grape_crush
> “Data integrity between multiple interfacing systems with varying formating rules is a pain in the ass.”
And how.
I’m guessing that they implemented a standard set of rules based on a sample data set, but when the actual data comes in, it invariably has exceptions that aren’t accounted for…and then you have rules conflicts that you have to watch out for so you don’t break something elsewhere.
And that doesn’t even begin to scratch whatever might be happening with the data handling on the insurance company’s side.
It’s not a simple thing by any means.
Fair Economist
It will get fixed eventually, but if there’s a 10% error rate with 3 weeks to showtime, there’s going to be some actual problems for the first month or two when people show up at the doctor’s with insurance not properly recorded.
pseudonymous in nc
Richard: is there any incentive to come up with an alternative standard interchange format, or are the parties involved just going to shrug and keep working around it by employing people to fix the errors?
Easier said than done, of course, but what healthcare.gov has exposed beyond its own internal code issues is that the ACA’s reforms are taking a private, smaller-scale B2B domain where formats and protocols just about scale to cope with the staff employed to deal with their messines, and exposing that domain to a messier public sphere. The intention was to be an evolutionary change by retaining the infrastructure of the private insurance and Medicaid, but that’s not always possible.
Thymezone
Refreshing to read the inputs of someone who understands the back end data problems. Good piece.
Yatsuno
@Richard Mayhew: Protip: that might be a valid SSN. I know you were trying to just make one up, but there is a possibility that is owned by someone.
eldorado
Data integrity between multiple interfacing systems with varying formating rules is a pain in the ass.
thirding this.
Richard Mayhew
@Yatsuno: Updated to deliberate gibberish
Richard Mayhew
@pseudonymous in nc: Not really, we get 834s from CMS and our state Medicaid agency on a daily/weekly basis with a consistent but very low error rate. That is because the CMS and state Medicaid people know what the hell they are doing and whenever they modify their systems, we know about it ahead of time and there is a three to six month “dummy” and “live data parraellel” testing regime before the switch is flipped.
I’m betting the developers or more importantly, the software architechts and project managers did not access the relevant 834 experts in a timely and consistent manner
RobertB
The unusual thing from where I sit was that they went to new 834s, from the original 5010X220 to the 5010X307. These became available relatively late, and much more quickly than the original 4010 => 5010 timelines were.
I’m definitely not an 834 expert – I’m much more knowledgeable about the X12 response documents than I am about the 834. But my wild-ass guess is that your 834 problems aren’t massive numbers of front-end errors, but are back-end issues cause by the new X307 being not quite identical to the older X220. Catching X12 syntax errors isn’t rocket surgery.
You can also make a case for Payers getting slammed by the 834s. Another wild-ass guess, but we’re looking at what, 10x the old 834 volumes? This isn’t real-time like 27Xs or anything, where you have to get a response in a very short time, but I bet that volume is a pain all by itself.
pseudonymous in nc
@Richard Mayhew:
I don’t doubt that. That said, there’s a broader question about how well that expertise scales, when you’re going from a system that’s built around a “trained operator” model to one that has to do some semi-intelligent data parsing. A system based upon people doing physical data entry can accommodate silent, experience-based edits to prevent simple encoding errors like an ampersand making its way into a text field, or handle more complex issues like a change of name, or an address that doesn’t normalize easily into the standard fields that the format demands. (I had to send a package earlier in the year to a friend who lives on a boat moored at a boathouse in England: the USPS address system is not very good at dealing with that.)
I suppose it’s a similar transition to the one that moved flight bookings online when they had been the domain of travel agencies.
LittlePig
Having written an EMC320 to ANSI837 converter for HCFA back in aught three, I feel their pain to an extent. There’s no excuse for missing the idiot errors, though – those were well-documented in the 837 anyway: e.g. must be alphanumeric, minimum length x. I would have thought they had substructuring issues; that’s the (somewhat) hairy part.
Bubba Dave
@LittlePig:
Late project, gotta slip something, and there’s this big bar on the Gantt chart labeled “Testing…” — You do IT, you know the drill. That’s the most frustrating thing about the whole debacle– there’s nothing new here, just traditional worst-practice project managment.
fuckwit
just signed up for blue shield silver 94 plan. i’m completely covered. i can’t afford it, but it’s cheaper than i could ever imagine. time to get all my broken shit fixed next year