Ssun Call 2 PM Thursday June 25
STD Surveillance Network (SSuN) Conference Call
SSuN Call #22; July 23, 2009
Participants
SSuN Sites
Alabama: Jane Schwebke, Elizabeth Turnipseed, Shuying Yu
Baltimore: Emily Erbelding, Katherine Harrison
Chicago: Irina Tabidze, Anna Hotton, Beau Gratzer
Colorado: Christie Mettenbrink, Melanie Mattson, Doug Robinson
Connecticut: Raul Pino
Los Angeles: Sarah Guerry, Michael Chien
California DH: Rain Mocello
Louisiana:
NYC: Preeti Pathela, Ellen Klinger
Philadelphia: Lenore Asbel, Greta Anchuetz, MelindaSalmon
San Francisco: Kyle Bernstein
Virginia: River Pugsley, Jeff Stover
Washington: Mark Stenger, Paul Swenson, Todd Rime
Minnesota (SSuN cycle 1): Summer Martins
CDC
Division of STD Prevention (DSTDP)
- Epidemiology and Surveillance Branch (ESB): Debbie Dowell, Lori Newman, Hillard Weinstock,
- Statistics & Data Management Branch (SDMB): Darlene Davis, Jim Braxton
- Health Services Research & Evaluation Branch (HSREB): Guoyu Tao, Charlotte Kent
Agenda:
1. Personnel update (Debbie)
2. GIS spatial analysis planning OPTIONAL call Thursday, July 30, 2-3pm EST (Debbie)
3. GIS Inventory Survey (Debbie)
4. Site visit updates (Debbie)
5. SSuN Collaborators’ meeting planning: Dec. 1st-3rd, 2009: save the date (Debbie)
6. Data management progress: (Darlene, Debbie)
7. Trich update (Debbie)
8. SSuN cycle 1 analyses (Lori)
9. Miscellaneous Items (anyone)
1. Personnel update (Debbie): we expect to have a new SSuN Project Coordinator to replace Shalini Parekh by early August and will send out her contact information once she’s here.
2. GIS spatial analysis planning OPTIONAL call Thursday, July 30, 2-3pm EST (Debbie): We will hold an OPTIONAL SSuN call Thursday, July 30thfor those who are interested in talking about geographic/spatial analyses using SSuN data. We are in the early stages of looking into best uses of geographic analyses of SSuN data, and welcome your ideas for analyses and priorities and your questions. Brian Lewis and Shannon Graham of the Geospatial Research, Analysis, & Services Program (GRASP) at ATSDR/CDC have offered us their expertise in mapping and spatial analysis for work on analyses of SSuN data and will participate in the call as consultants.
3. GIS Inventory Survey (Debbie): Our Colleagues in CDC’s “Evaluating Integration of HIV/AIDS Surveillance Data with a Geographic Information System" Project(which includes 3 sites also participating in SSuN, Colorado, Virginia, and Washington), have been developing a Policies and Procedures Manual for health departments to use for guidance in integration of GIS with HIV surveillance data and have asked for our help in making sure the manual makes sense to health department staff at all levels of GIS experience, including no GIS experience. They’d like 3 volunteers from SSuN sites (not including those 3 SSuN sites already on the project) to review the manual and give feedback. Participation in this is NOT a required SSuN activity, but might be interesting, especially if your site is considering doing more work with GIS. Does anyone object to receiving an email with more information about this? [No objections].You’ll receive an email about this in the next week or so.
4. Site visit updates (Debbie): We have several upcoming SSuN site visits:
Connecticut-August 31st (Debbie, Hillard, Jim)
Chicago- September 16th (Niko, Hillard, Darlene)
Baltimore-September 24th (Debbie, Deblina, Darlene)
5. SSuN Collaborators’ meeting planning: Dec. 1st-3rd, 2009: save the date (Debbie): thanks everyone for sending in the dates that would work for the SSuN Collaborators’ meeting. It seems that the dates that will work best for everyone are Tuesday-Thursday December 1st-3rd. So please save those dates for a trip to Atlanta! We may end up only needing 2 of those 3 days and will keep you posted.
6. Data management progress: (Darlene, Debbie)
Test clinic data: We’ve received test clinic data from 6 sites: 3 sites have sent all 3 files for the clinic data (that is standardized and core as one file, and also lab and diagnosis files. 3 additional sites have sent core and standardized clinic test data. We’d like to receive test clinic data from all sites by September 15th(including core and standardized data as 1 file, lab data as another file, and diagnostic data as another file). Please let me and/or Darlene know if sending test data by September 15th will be a problem at your site.
Test county data: We’ve received county test data from 2 sites.
We’d like to receive test county data from all sites by October 15th. Please let me and/or Darlene know if sending test county data by October 15th will be a problem at your site.
Question (Kyle): have there been any issues with the test data so far?
Answer (Darlene): Only minor issues that were easily resolved.
We’ve had a couple questions recently about the countydataset and wanted to share these and discussthem:
1. (Debbie): Some sites asked about sending morbidity data for cases not interviewed. Yes, you should send data for all GC morbidity in the jurisdictions for which you are sending SSuN data. Those cases you successfully interview shouldbe embedded in the samefile with allthe morbidity data. All cases in the morbidity file should have a code for whether or not eligible for SSuN; if eligible, should have a random number assigned anda code for whether they were selected for interview; and ifselected, should have an interview status code. Generally, successfully interviewed cases will have more data elements completed, but we want whatever data elements you have, limited though they might be, whether or not the interview was completed.
(Lori): Your site’s morbidity data is more complete and higher quality than the data we get through NETSS, and the SSuN cycle 1 morbidity data, even for the non-interviewed cases and limited though they might be, have proven very useful in Debbie’s Interrupted Time Series treatment analysis and in my Access to Care analysis. We would not have been able to get as high quality provider type or treatment data through NETSS.
2. (Debbie):Sampling fraction threshold and randomization number: we’ve had a couple questions on this recently. Ideally, the sample fraction should be set to give a good estimate of interviews needed for a total of 20 interviews successfully completed/month. This might be difficult to estimate, especially at first. We can help by providing ranges from other sites in cycle 1. Let us know if that would be useful. Once someone is selected for interview (by having a random number under the threshold for sample fraction), you should attempt to interview them, even if you've reached your quota for the month. If you are consistently successfully interviewing more than you need, the sample fraction should be reduced. Most sites are generally adjusting sample fraction quarterly, but it could be adjusted more frequently, up to once/month if needed. So there should not be "leftover" cases, because anyone who has a random number under the sample fraction should have an interview attempted.
(Lori): We don’t expect everyone to get exactly 240 cases per year.
Question (Mark): In WA, some jurisdictions have different sampling fractions; is this acceptable?
Answer (Debbie): Yes, as long as we know what the sampling fraction is for each case.
Answer (Lori): The “SAMPFRACT” variable is used for this.
Question (Rain): We have “too much” treatment data in CA (multiple treatments are listed) and are not sure how to fit them into the SSuN data element fields.
Answer (Mark): We had this issue in WA as well and have dealt with it by collecting all treatments given at the same time, then entering the 1st treatment listed which is specific for gonorrhea as Treatment 1, and the 1st treatment listed which is specific for chlamydia as Treatment 2.
Answer (Debbie): WA’s approach sounds very reasonable, and Rain, we can also talk more about this offline for any remaining questions.
Fluoroquinolone treatment data: We’ve received fluoroquinolone treatment data from 3 sites. This was NOT a required SSuN activity, but we think will it be useful in looking at fluoroquinolone prescribing in general, and useful for the participating sites. Thank you to Connecticut, Alabama, and Baltimore for sending those data. Louisiana plans to look into these data as well and send them. We’ve done some preliminary analysis of data completeness, and I’ll be in touch with the individual sites soon with these preliminary results and with some questions.
7. Trich update (Debbie): Bob Kirkcaldy reports that five out of six sites have now sent isolates. So far, we’ve received 80 specimens, and 64 (80%) have been successfully assayed for drug resistance.
Three of the 80 have shown low level resistance of around MLC= 50; the rest have been lower.
8. SSuN cycle 1 analyses (Lori): We wanted to spend some time on this call talking about the most important analyses to complete and messages to get out from SSuN Cycle 1. What we learn from SSuN Cycle 1 will help with SSuN Cycle 2. You can refer to the attachments sent out with the call notes: My ISSTDR poster on Access to Care, Bob's abstract from ISSTDR on HIV/GC co-infection in MSM, and the list of completed and ongoing projects. Did anyone have anything to add to the list? Does anyone have thoughts about what were the important main messages to get out about SSuN cycle 1?
(Charlotte: ) Tracking and monitoring of cases diagnosed outside of STD clinics over time is important
(Lori: )The Access to Care analysis,presented as a poster at ISSTDR, used the Countydataset from SSuN Cycle 1, used cleaned up/beefed up case report data from each of the sites geocoded down to the census tract level. We’re using the proportion diagnosed in ER’s as a proxy for not optimal care—patients don’t always get appropriate treatment, it costs more, they may not get HIV prevention services, and it’s not a great environment for partner management. We could also look at this in terms of how we can improve care in ERs. Our results by provider type showed that women, younger patients (both 15-19 and 20-29), blacks, and those living in a census tract with >=20% below poverty were more likely to present in ERs (see table and figures). Also, ERs accounted for 43.9% of cases diagnosed on weekends and 13.7% of cases on weekdays.
I’m cautious about using multivariate analysis to look further at this because the results are from 5 very different sites and there are still important differences in how sites code data.
Do we want to turn this into a paper? The idea of access to care is really timely, important, and interesting, and is being discussed at all levels of government. One of the aspects of SSuN that makes it unique is the ability to explore this issue.
(Charlotte: ) Important to have a baseline measure. If treated in Primary Care, are they following guidelines better? Important 1st paper and great data to publish over time.
Issues with provider type classification:
(Mark: ) It would be very useful to spend some time as a group refining definitions of provider type. There are issues, e.g., in urgent vs. emergency settings.
(Lori: ) Community health centers are another important site variable. There is a code (PROVIDER=18) in SSuN Cycle 2 data elements, which is “Public Clinic, (not STD)/CommunityHealthCenter” which was our attempt to capture this.
(Charlotte: ) They are safety net providers and will probably account for the largest expansion in the next 5 years with the most new money.
(Mark: ) WA has tried to recruit as many Community Health Centers as possible into IPP to have good chlamydia data.
(Emily: ) these data can be difficult, we might get provider/address but not facility
(NYC: ) our reports come as a mixed bag as well, and we’re often not sure where they fit for provider type
(Summer: ) ER and Urgent Care are currently lumped together. We realized that we included provider type during the interview as an oversight. When we looked at primary care visits on the case report form, according to patient report 27% were actually urgent care visits.
(Lori: ) It would be great if you could send us that information as it would help us describe some of these limitations of our provider type information.
(Debbie: ) It would be very useful in describing limitations in the fluoroquinolone treatment analysis as well.
(Raul: ) I realize we have also just been asking and not collecting provider type from the case report—will go back and like MN, compare case report with patient report of provider type.
(Lori: ) There is also “other hospital.” This was an issue in Colorado. We may be underestimating emergency medical facilities.
(Charlotte: ) A lot of people who go to urgent care are different from those who go to the ER; urgent care visit doesn’t necessarily suggest they don’t have a “medical home.”
(Rain: ) there are a number of fee-for-service urgent care clinics in CA. There is a big difference in prescribing patterns: urgent care clinics are worse than ERs in inappropriate treatment (e.g. fluoroquinolones). Not a difference in hospital-affiliated vs. urgent care.
(Lori: ) we won’t be able to resolve these provider type issues on this call. But may be worth discussing further in the future.
Issues with collecting insurance status:
(Raul: ) we are not collecting information about health insurance. I will add this on for my interviews.
(Lori: ) SSuN considered looking at insurance status, but many sites felt this information to be confidential and hard to collect in a verifiable and useful manner. Instead, SSuN has chosen to look at insurance status through ecologic analysis: e.g. by census tract insurance coverage in a patient’s residence census tract.
Mapping issues:
(Lori): started working with Brian Lewis to map cases. Have to make at least 5 maps because not contiguous areas; they are all so unique geographically. Started mapping simply—cases that attended an ER (chloropleth) with points where ERs are; STD cases and STD clinics. Did with cases, then with rates.
(Mark: ) we looked at distance using a Winsorized mean for each census tract, stratified by provider type, and looked at GC rates by quintile at census tract level. Plan to use both straight line and network distance.
(Jeff: ) Before our spatial call, could someone find out whether you have access to Claritas market segmentation data at CDC? This is different information than what we get from SSuN, down to the household level, and could be useful.
(Debbie: ) will find out about this.
Issues with looking at symptomatic patients:
(Charlotte: ) If 44% diagnosed on weekend, suggests there isn’t alternate care on weekend. Do you have data on symptoms?
(Lori: ) We only know about symptoms in patients who were interviewed.
(Debbie: ) we also looked with Mark Stenger at WA data for delay to care for symptomatic males (could only do this for interviewed patients but WA interviews more patients than required for SSuN) and found one of the most important predictors for delay of care for more than 3 days was development of symptoms on a Thursday.
(Kyle: ) Some urgent care/ERs may have screening policies. Would limit to those who are symptomatic.
(Summer: ) our routine surveillance has information on symptoms.
(Mark: ) we also have this in WA.
(Lori: ) this is why, even though it may not be clear in advance, it may be useful to have data from your morbidity files on non-interviewed patients (because you may have more complete and higher quality information on symptoms, provider type, and treatment than we could obtain through NETSS).
9. Miscellaneous:
- Optional GIS call Thursday, July 30th, 2-3pm EST
- Next SSuN call Thursday, August 20th, 2-3pm EST