MCHB/EPI Miami Conference — December 7 - 9, 2005
Linking Data and Policy to Programs — Transcript
HEIDI BROWN: So the title of my talk is a little bit different. I apologize for that.
Good afternoon. I'm really excited to share with you all some results of a study I did last year at CDC. I'd just like to acknowledge my mentor at CDC on the project, Stephanie Schrag, who is being awarded at this conference although she is not here yet in NCID and the Respiratory Diseases Branch at CDC. Also my co-collaborator, Matt Avery at the North Carolina State Center for Health Statistics, he is the North Carolina PRAMS coordinator.
So I'm really excited to share these results with you and I'm going to answer three questions, hopefully, which are: one, what is Group B Strep; two, how are we doing with universal screening, in other words, how many women are being screened and who are they and how can we do it better?
So by way of background streptococcus a glaciate, group B Streptococcus or GBS is the leading infectious cause of neonatal morbidity and mortality. There are actually two forms of neonatal group B strep: early onset, which occurs in the first week of life, and late onset, which occurs at days 7 to 90. Early onset group B strep is transmitted from mother to child during delivery where as late onset GBS is acquired in the neonatal period. Both forms of neonatal GBS result in sepsis pneumonia and meningitis and this is an overview of how the transmission tree breaks down.
So approximately 10 to 35 percent of pregnant women are colonized with GBS. Of those women, 50 percent will go on to pass the bacteria to their newborns who will be colonized. Of colonized newborns, 98 percent will be asymptomatic but two percent will be very sick babies with the early onset disease including sepsis, pneumonia or meningitis. Of those babies, five percent will die and an unknown percentage may have long-term neurologic sequela.
So it is not a small problem. But the good news is that we can prevent perinatal transmission and the way we do it is by administering intra partum antibiotics to moms and this can prevent colonization of newborns and all of those nasty sequela.
So who should get antibiotics? In 1996 a consensus statement recommended two equally acceptable strategies. Doctors could either do cultures on all pregnant women as GBS screening and then those women who were GBS positive would receive antibiotics at delivery. Or they could give prophylactic antibiotics to any woman with defined risks factors for either GBS colonization or transmission and they didn't have to screen those women. But in 2002 there was a landmark study published in the New England Journal of Medicine that found that screening was greater than 50 percent more effective than the risk based strategy for intra-partum prophylactic. So in 2002, CDC in conjunction with the professional organizations for pediatricians and obstetrician and gynecologist issued an updated guideline and a consensus statement recommending universal culture-based screening for vaginal and rectal GBS colonization at 35 to 37 weeks gestation.
So the objectives of this study were to look at pregnant women in North Carolina during 2002 and 2003, right around the time of the policy change, to examine rates of reported GBS screening and knowledge of whether or not women had been screened, as well as to identify risk factors for not being screened for GBS or for not knowing whether or not they have been screened.
So some background on PRAMS, which I'm sure, most of you are familiar with. The Pregnancy Risk Assessment Monitoring System or PRAMS is a monthly mail and telephone survey targeting a population-based random stratified sample of women who have recently delivered a live born infant. It includes a core questionnaire, which all states participating in PRAMS ask, and it also has standard questions, which states may opt to include in their questionnaire.
In 2002 and 2003 there were three GBS questions in the standard and optional component that states could chose to ask. One asking about awareness of GBS. One, talking about whether or not a health care provider discussed GBS with the patient and then one asking about whether or not women have been screened. PRAMS survey data are linked to birth certificate data and are weighted in a complex way to account for sample design, nonresponse and noncoverage. States need to attain at least a 70 percent response rate in order for their data to be useable which is really impressive to attain a 70 percent response rate. There is a two-year lag between data collection and availability.
So these are the states that participated in PRAMS during 2002 and 2003 and actually, both New York and North Carolina asked some of the GBS questions but only North Carolina obtained a 70 percent response rate so that is the data that I used.
So I looked at the years for 2002 and 2003 separately first and because the numbers were pretty similar I combined the data. The real outcome that I was interested in was the response to this question: "At any time during your most recent pregnancy did you get tested for the bacteria Group B Strep or Beta Strep?" And women could answer: yes, no or I don't know.
When we first looked at this, it appeared that women who answered yes were different from women who answered no, and women who answered I don't know were different from women who answered yes or no. So we decided to actually do two models rather than just excluding the I don't knows or collapsing them with the no's. First we did a model looking among those women who are screened, who know their screening status I'm sorry. Among those women who know whether or not they've been screened. Looking at women who said no and women who said yes. Then we did another model looking at those women who didn't know whether or not they'd been screened and comparing them to the policy goal of women who said yes they had been screened.
So first we did invariable analysis and any variable of interest that had a P value less than .2 was included in our initial multi-variable models. Then we used backward stepwise logistic regression to do our final models and main affects with a P less than 0.05 were included in the final models.
So here is what we found. Approximately 236,000 live births in North Carolina during 2002 and 2003. About 4,100 women were invited to participate in the PRAMS survey and 3,000 responded for a response rate of 73 percent; the majority of the sample self-identified as white and 12 percent self-identified as Hispanic ethnicity. About half had their delivery paid for by Medicaid. The majority or two-thirds of women reported their primary source of prenatal care as a private doctor or a health maintenance organization and about one-third of women went to a hospital or a health department clinic. Less than one percent received no pre-natal care.
So in response to our question of interest, 72 percent reported having been screened for GBS. Nine percent reported not having been screened, and 19 percent did not know whether or not they had been screened. And just to give you a little bit of comparison information, 82 percent reported that they had received HIV testing during pregnancy and 82 percent reported that a health care worker discussed Group B strep with them.
So this is the results of our invariable analysis of the first model. This is among those women who knew their screening status, yes versus no. The things that were correlated with not being tested for GBS were younger age, lower education, Hispanic ethnicity, being unmarried, having delivery paid for by Medicaid, seeking pre-natal care at a hospital or health department clinic as opposed to from a private doctor or HMO, not having insurance before pregnancy and not having pre-natal HIV testing. In our multi-variable model, three factors remain significant. Those are highlighted in gold and we did control for gestational age to try to exclude women who delivered prior to the 35 weeks at which screening is recommended. You can see that Hispanic ethnicity, seeking prenatal care primarily at a hospital or health department clinic or not receiving HIV testing during pregnancy were correlated with not being screened for GBS.
This is the result of the second model comparing those women who did not know whether or not they had been screened with those women who definitely said they had been screened. You see the things above the white line represent the same factors you saw in the no vs. yes but then in addition you see race other than white, so women who self-identified as black or other race, those women whose pregnancies were unintended and those women who received WIC benefits during pregnancy were less likely to know whether or not they had received GBS testing.
In the multi-variable model again we saw Hispanic ethnicity, prenatal care at a hospital or health department clinic and lack of prenatal HIV testing but we additionally saw a delivery paid for by Medicaid and race other than white as significant risk-factors for not knowing whether or not women had been screened for GBS.
So in summary, among women participating in North Carolina PRAMS in 2002 and 2003, 72 percent reported being screened for GBS during pregnancy and risk-factors for lack of screening included Hispanic ethnicity, prenatal care from a hospital or health department clinic and lack of prenatal HIV testing.
Nineteen percent did not know their GBS screening status and risk factors for lack of knowledge included those three in addition to black race or other race and delivery paid for by Medicaid.
There are at least four limitations to this study. First of all, it is a self-reported survey so we can't confirm GBS screening status. We can only comment on reported rates of screening. One thing that I do think is valuable about this however is the proportion of women who said I don't know. Because those women, regardless of whether or not they were screened, they didn't know that. So if they presented for example to a different hospital than where they received prenatal care and their chart wasn't available, they wouldn't know to ask for antibiotics or to say, "Oh no, my doctor said I don't need antibiotics. I don't have GBS." The second limitation is that health care providers might not use the same wording as the PRAMS question so maybe the doctor just said, "Hey, when you go deliver, just tell them you don't need antibiotics," and didn't mention GBS so the women didn't recognize that in the question. Third limitation is that we can't generalize these findings to the entire country since we are just looking at North Carolina . But the good news is that in the 2005-2006 PRAMS data eleven states adopted these questions so hopefully we'll be able to see a little bit broader cross section. And lastly, there is no base line data with which to compare these screenings rate. Although we do know that the national average before 2002 was about 50 percent. It is entirely possible that North Carolina has always been above the national average in terms of screening but that's the only baseline we have.
So what does it all mean? In pink; yeah, okay. You can see the colors right? In pink you see the incidents of early onset GBS disease and you see it steadily going down and actually just this week in MMWR they published the most recent numbers for 2004 and the incidents is 0.34 per thousand live births. So it is going up a little bit but still staying in a downward trend.
In green is the Healthy people 2010 goal for the early onset disease and you can see that we are well below it over all. However it is important to note that black incidents are actually above this goal. So this is white women driving this sample of looking like we are meeting the goal because we are not for everyone.
In yellow is just late onset disease and you can see it is a relatively stable curve because it is not affected by the prenatal screening. This first downward slope is in 1996 when the first consensus guidelines came out and this little divot is in 2002 and we are hoping in future years we will see an increasing downward slope but we'll have to wait and see.
So for my discussion since the title of this session is linking data in policy to programs, I wanted to sort of talk about what all this means and how we can do a better job preventing GBS. Before 2002 when providers could either just do a risk-based intra-partum prophylactic or they could do a universal screening it didn't really make sense to target the general public with information because providers might not be doing routine screening. But since 2002 we now have the opportunity to really do a two-pronged approach. We can encourage providers to do the screening and we can also educate women, talk to your doctor. Ask about GBS screening. Protect your baby. Know your GBS status.
What can we do in North Carolina ? I think the most striking finding for me was among Hispanic women and I didn't actually present this number but among Hispanic women, the proportion that didn't know whether or not they had been screened was also 50 percent. Among non-Hispanic women it was 16 percent. Maybe some of that is due to language barriers because the questionnaire administered in Spanish and maybe some prenatal care providers aren't fluent in Spanish. But still, we are not doing a good job educating this population regardless of the reason behind it. So we need to do better. We also need to target women who miss other recommended prenatal interventions like HIV testing. We sort of saw that variable as sort of proxy for adequacy of prenatal care.
Lastly, women seeking care in hospital and health department clinics. I want to be really careful not to imply that hospital and health department clinics in North Carolina aren't doing this screening because that is not what I think this is showing. I think that probably they have a different population who aren't presenting to care regularly. Perhaps they are too busy to talk about it and so I don't want to draw any conclusions but that is something the North Carolina folks are going to explore a little bit more.
And this is just newly available on the Web site this week from my branch at CDC. This is our GBS pamphlet for the general public. It is available in English too but I just thought I'd use Spanish because of my Hispanic women finding. It basically says, "Are you pregnant? Protect your baby, talk to your doctor, find out your status. You don't need to get treated now but at delivery you might need antibiotics." It is very user friendly and you can actually get it is you want to use it in your state. The brochures are available in both English and Spanish and you go to our Web site or you can call CDC is you want limited; limited means like less than 15 copies. So if you want more then you have to call the number for bulk orders.
Lastly, I would just like to acknowledge my colleagues at North Carolina and at CDC.