Archived Comments for:
Reduction of late stillbirth with the introduction of fetal movement information and guidelines – a clinical quality improvement
Education/information may be necessary to decreased fetal movement especially “at term”.
Shigeki Matsubara, Department of Obstetrics and Gynecology, Jichi Medical University, Tochigi, Japan
24 August 2009
Dear Sir, Although maternally perceived decreased fetal movement (DFM) sometimes precedes imminent fetal jeopardy, controversy remains as to whether universal screening, ie,, informing all pregnant women of fetal movement (FM), and identifying women with DFM followed by intervention, reduces the stillbirth rate.
We applaud Dr Tveit and her colleagues for having shown that universal screening for DFM did reduce the stillbirth rate in eastern Norway [1]. A well designed brochure [2] accompanying this article is applicable to any other institutes or countries.
Recently, we have introduced “modified count to 10” for FM, and established its reference value for low-risk Japanese [3]. As is commonly known, perceived FM decreases toward term, with the 32th week showing the highest rates of FM [3].
More recently, we encountered a clinically significant case. A 26-year-old Japanese primigravida visited us for a regular pregnancy checkup at 39+0 weeks. She informed us of DFM over the last 3 days. Unfortunately, the fetus had already died. She had recorded a “modified count to 10” chart from the 27th week. She was previously educated/informed that DFM was sign of a fetal jeopardy and that, in cases of DFM, she should consult us immediately. Unfortunately, she was also notified that FM may decrease toward term: she thought that DFM this time may be due to her being near term and preceding signs of labor onset. That was why she did not consult us immediately, and waited 3 days.
Based on Tveit et al.’s article [1] and our experience, we emphasize the following 3 points. First, it is true that FM generally decreases toward term, but this should not be over-emphasized as it may mislead pregnant women. Second, pregnant women should be encouraged to consult a care provider whenever experiencing DFM irrespective of the number of gestational weeks, even when near the expected due date. Tveit et al. [1] also touched on this issue, but did not emphasize DFM near term. Third, the point mentioned above should definitely be stated in the brochure [2]. In the brochure, it is described: “even as you approach the expected date of delivery, your baby will seldom sleep more than an hour at a time” [2]. It may be better to more clearly describe this: “although FM may decrease near term, it is a subtle and slow decrease. Whenever you subjectively feel a movement decrease, contact us immediately, even near the expected due date” Such a description may be added in the next revision of the brochure, which may markedly improve it.
References 1 Tveit JVH, Saastad E, Stray-Pederson B, Bordahl PE, Flenady V, Fretts R, et al.: Reduction of late stillbirth with the introduction of fetal movement information and guidelines –a clinical quality improvement. BMC Pregnancy and Childbirth 2009, 9:32 doi:10.1186/1471-2393-9-32 (http://www.bomedcentral.com/1471-2393/9/32 (accessed on August 20, 2009) 2 http://www.biomedcentral.com/imedia/1876556089277830/supp2.pdf (accessed on August 20, 2009) 3 Kuwata T, Matsubara S, Ohkusa T, Ohkuchi A, Izumi A, Watanabe T, et al.: Establlishing reference value for the frequency of fetal movements using modified “count to 10” method. J Obstet Gynaecol Res 2008, 34:318-23.
Correspondence: Shigeki Matsubara, MD. Professor of Department of Obstetrics and Gynecology of Jichi Medical University, 3311-1 Shimotsuke, Tochigi 329-0498, Japan Tel.:+81-285-58-7376 Fax: +81-285-44-8505 E-mail: matsushi@jichi.ac.jp
Competing interests
None
Reduction of late stillbirths - questions to the authors
Kjell Salvesen, Trondheim University Hospital
15 October 2009
I have read this paper with interest. An intervention reducing the stillbirth rate by 50% in a group of women presenting with decreased fetal movements (DFM), and 30% in the study population as a whole, is good news. However, I would like to ask the authors some questions regarding their results.
In brief, the study was a comparison of obstetric outcomes in two time periods: baseline (April – October 2005) and during an intervention period (November 2005 – March 2007). All women delivered at 14 maternity wards in Norway. The intervention consisted of: 1. An informational brochure on fetal movements (including a kick chart) was distributed to pregnant women attending the routine ultrasound scan at 18 weeks of pregnancy. 2. New guidelines for health care workers on taking care of women presenting with DFM were implemented. The guidelines recommended: “a standard clinical evaluation for all women reporting DFM, an NST, and an ultrasound scan to quantify FM, amniotic fluid volume and fetal anatomy and growth. A mother presenting with a concern of DFM was to be examined within 2 hours if absence of FM was suspected, otherwise within 12 hours”.
The authors present a reduction in stillbirth rate after 27 weeks of gestation, but no change in the proportion of women presenting with DFM (6.3% vs 6.6%, p= 0.19).
Baseline (April – October 2005): There were 19 407 births, 50 stillbirths, and 1215 consultations for DFM. The stillbirth rate among women with DFM was 4.2% (50/1215). Stillbirth rate in the baseline population was 3.0/1000. Total number was not reported in the paper, but can be calculated to 58 stillbirths - 50 reported in the paper and 8 not reported in the paper, presumably among women without DFM.
During the intervention (November 2005 – March 2007): There were 46 143 births, 73 stillbirths and 3038 consultations for DFM. Stillbirth rate among women with DFM was 2.4% (73/3038). Stillbirth rate in the population exposed to the intervention was 2.0/1000. From this information I can estimate 92 stillbirths - 73 reported in the paper and 19 not reported, presumably among women without DFM. The effect on stillbirth rates was observed immediately after the intervention was implemented. Figure 2 in the paper demonstrates an abrupt downward shift in the curve from November 2005 onwards, and a significant change in stillbirth rate was observed after 7 months.
I kindly ask the authors to clarify how the change in the stillbirth rate from November 2005 onwards can be explained by the intervention? 1. Registration of FM was advocated in the third trimester (after 28 weeks). Women receiving an information brochure at 18 weeks will not come for DFM or fetal death until several months later. Hence, information to women in mid-pregnancy cannot explain the abrupt decline in stillbirth rate after study start. 2. There was no change in the proportion of pregnant women presenting with DFM (6.3% vs 6.6%, p= 0.19). It is therefore unlikely that the intervention have increased the sensitivity of counting FM as a test for fetal death risk. 3. The new guidelines included CTG and ultrasound (biophysical profile, fetal growth, Doppler). Table 1 demonstrates a marginal increase in the use of CTG (96% vs. 98%) and ultrasound (86% vs. 94%) among women presenting with DFM, but no change in the use of Doppler. These fetal surveillance tests have not reduced perinatal mortality in previous randomised controlled trials - even in high-risk pregnancies, and it is well known that the tests have poor sensitivity and specificity. It is unlikely that a small increase in the use of CTG and ultrasound can explain a reduction in the stillbirth rate by 50%.
Stillbirth rates in Norway 2005-06 The paper states: “In addition to the registrations by our study protocol, the numbers of births and stillbirths from our population were obtained from the Medical birth registry in Norway to assess overall trends in still birth, for the most updated period available: April 2005 to December 2006”.
Stillbirth rates from the Medical Birth Registry of Norway are accessible on the internet (www.mfr.no). One cannot access hospital specific rates from the ”14 delivery units in eastern Norway and the city of Bergen”. However, it is possible to analyze stillbirth rates from 8 counties in eastern Norway (Oslo, Akershus, Buskerud, Vestfold, Østfold, Telemark, Hedmark, Oppland) and Hordaland (including the city of Bergen). The deliveries as recorded in www.mfr.no in these 9 counties are, as far as I know, equivalent to the deliveries at the 14 delivery units in the study.
2005 Eastern Norway + Hordaland N = 33 981 births. N = 70 stillbirths. Rate: 2.1/1000 Other counties N = 22 859 births. N = 72 stillbirths. Rate: 3.1/1000 Total in Norway N = 56 876 births. N = 142 stillbirths. Rate: 2.5/1000
2006 Eastern Norway + Hordaland N = 35 227 births. N = 85 stillbirths. Rate: 2.4/1000 Other counties N = 23 384 births. N = 66 stillbirths. Rate: 2.8/1000 Total in Norway N = 58 611 births. N = 151 stillbirths. Rate: 2.6/1000
I kindly ask the authors to clarify why the study reports a 30% reduction in the population, when data from the Medical Birth Registry of Norway shows no reduction in the stillbirth rate in the counties included in the study during 2005-06. It would be helpful if the authors report the number of stillbirths in 2005 and 2006 among women: 1. with DFM in the 14 delivery units 2. without DFM in the 14 delivery units 3. who did not deliver in the 14 delivery units, but in the 9 counties
Reduced rates of preterm births and inductions of labour after intervention The authors report a borderline statistically significant reduction in preterm births (12% vs. 10%, p = 0.1) and a statistically significant reduction in admissions for induction of labour (7.0% vs. 4.9%, p = 0.03) after a multivariate analysis. In addition, they report a small, non-significant reduction in admissions for emergency section (1.8% vs. 1.2%, ns.).
An effective intervention that reduces the stillbirth rate by 30-50% is likely to be accompanied by an increase in the number of induced deliveries, the numbers of preterm births and Cesarean sections. This study reports a statistical significant reduction of induced deliveries and preterm births during the intervention. I kindly ask the authors to clarify how a reduced stillbirth rate can be accomplished by a reduction in preterm births and in inductions of labour.
Kind regards
Kjell Å. Salvesen Professor National Center for Fetal Medicine Trondheim University Hospital N 7006 Trondheim
Competing interests
None declared
Reduction of late stillbirth - reply from the authors
J. Frederik Frøen, Norwegian Institute of Public Health, Oslo, Norway
29 October 2009
Dear Sir,
Salvesen’s comment to our study has just come to our attention, and we appreciate the opportunity to clarify. The comments have questions regarding three aspects of our study: 1. How come the effect is not reflected in publicly available data? 2. How come mortality seems lower already in the first month of intervention? 3. How come the intervention seems to lower mortality without increasing maternal concern and/or interventions?
1) To the first question, on why the effects cannot be seen on the web site of the Medical Birth Registry of Norway (MBRN), we regret that Salvesen has had the misfortune of using incorrect 1) Inclusion criteria, 2) Catchment criteria, 3) Catchment area, 4) Time period, 5) Intervention period, and 6) Gestational age. 1. The inclusion of pregnancies differs as our study only includes singleton pregnancies, while the MBRN web site Salvesen refers to does not stratify by pluralities and different mortality rates. 2. The catchment criteria differ. Salvesen interprets data from the MBRN web site as births in nine counties, but as specified in the first column of the tables the MBRN web site tables are not based on place of delivery, neither by county nor by institution, but on the legally registered maternal residential address. Norwegian women may attend any hospital of their choice free of charge, and frequently choose to receive their care for pregnancy and childbirth in another county if this hospital is geographically closer. The comparison with our data based on place of delivery is thus inaccurate, and further so simply because many young women live elsewhere than at her registered residential address, e.g. students. 3. The catchment area differ, as even if it was correctly based on the place of delivery, as discussed above, the simplified tables on the web site are based on counties, while our study is not. Salvesen states that ”The deliveries (…) in these 9 counties are, as far as I know, equivalent to the deliveries at the 14 delivery units in the study.”. This is unfortunate, as the MBRN web site he extracts data from lists and presents births from 22 institutions in these nine counties. 4. The time period is incorrect as these annual figures presented on the web site are incomparable with our study, and in particular the pre-intervention data are severely misrepresented. The figures presented by Salvesen add five non-included months to the seven pre-intervention months, and remove five included months from the intervention. 5. The intervention period is incorrect, adding to the inadequate time period above, as two months of data from the intervention period is moved into the pre-intervention period. 6. The gestational age may differ as many stillbirths (approximately 0.5 per 1000 births) are reported with unknown gestation at the time of reporting to the MBRN, while this may be established at a later date. In our material all stillbirths were dated after validation by the individual participating hospital. Unfortunately, such comparisons of incompatible data sets do not add helpful information. To do the comparison, well-selected data sets from the MBRN must be used, and these are publicly available to any interested scientist through the web site BMC. While preliminary numbers for 2007 are now available on the web site, the validated data for research purposes have not yet been released, but we will, of course, take keen interest in a new dataset to compare with our own data collection. We would also recommend not to base the comparison solely on the registered gestational age, but use all singleton stillbirths reported with a gestational age > 28 weeks, or unknown gestation, in combination with the WHO criteria of 1000 grams or 35 cm body length. This would avoid loosing the data due to unknown gestation and also avoid cases of early fetal death delivered several weeks later. Our agreement with the individual hospitals is, however, that no hospital-specific data will be published.
2) Second, how the “change in the stillbirth rate from November 2005” could be explained. The presented figure is a statistical process control (SPC) chart, a time series tool of simplified statistics much used in clinical quality improvement to help understand variation and change over time. SPC provides simple rules to identify change without misinterpreting change from one measurement to the next as a “change”, but normal variation in the observed process. Thus, as noted in the legend and marked with an arrow on the chart, we make no claim that there was any change on the SPC chart before June 2006. The “change” in November 2005 might simply be normal variation, as none of the measurements are outside the control limits marked with dotted lines. That said, we do see that the chart looks intriguing already from the very start of the intervention. Salvesen correctly notes that the information for women was distributed at approximately 18 weeks of gestation, but the description of “Registration of FM was advocated in the third trimester(…)” as the “active ingredient” of this campaign is inaccurate. Reducing this information campaign to a third trimester kick counting procedure passes the point. The purpose was to provide knowledge and information to pregnant women on how to recognize the difference between normal variation in healthy movements and worrisome signs, as well as some tools to support her decision making. Such campaigns typically create substantial attention initially and then loose the strength of novelty (in accordance with the SPC chart of this project). The launch of the intervention included not only handing out information to the first few women, but all midwives, community midwives and nurses, general practitioners, obstetricians, delivery units, antenatal care clinics, etc, in our catchment area were contacted with information in writing and/or with visits and meetings. There were news stories in local newspapers and on radio programs, pregnancy-related web sites linked to our project web site and sparked discussions in their forums where pregnant women at all gestations participated. If at any time there was a historically high awareness towards fetal activity in the catchment area, it would have been during the early phase of our intervention. Identifying the exact portion of such an information campaign that “explains” the effects seen is obviously difficult. We have, however, submitted a secondary paper to this study that deals specifically with how the campaign affected maternal attitudes and health seeking behaviour in different high and low risk segments of our population, and hope this will bring additional clarification when published. Also, as discussed more below, the information campaign was only one of two integral pieces of this quality improvement project, as the new guidelines for professional management were also implemented at the same time.
3) The third group of questions (2 & 3 and the final two sections) ask us to discuss further how an “information and management bundle” like our intervention can reduce adverse outcomes with apparently no downsides. The first notion that no increase of women presenting with DFM is equivalent to no improvement in “sensitivity of counting FM as a test for fetal death risk”, is mistaken in two aspects: Firstly, it is a misconception of the term sensitivity. Increasing the total number of test positives has no implications for the sensitivity of the test. This will depend on whether the test was positive in a sick or healthy pregnancy, as the sensitivity is defined as the proportion of sick pregnancies in the population that are captured by the test. Secondly, we have clearly had no wish to increase the proportion of women reporting DFM, and the inaccurate notion noted above, that the intervention was “kick counting”, plays a role also for this question. The purpose of the information provided both to pregnant women and their care takers aimed to increase the number of women reporting DFM when they were rightly concerned for DFM, while reducing the number of reports of DFM from women who were probably unneededly concerned for DFM. This is what would change sensitivity, and we do not claim that “kick counting” is what provides improved sensitivity. In terms of whether a small increase of fetal surveillance with CTG, ultrasound and Doppler (the latter not a part of our intervention) per se would explain all the reduction in stillbirths, we clearly agree with Salvesen that this would be highly unlikely. We also agree that most tests “have poor sensitivity and specificity”, but Salvesen’s generalized claim that none are effective in high risk pregnancies seems incompatible with evidence. Yet, the general merit of fetal surveillance is beyond this discussion and the question is rather how to improve its predictive value. As discussed above, what drives such change (given unchanged technical abilities of the test) is the proportion of sick pregnancies among those being tested. We would therefore suggest not to consider the correlation between outcomes and test prevalence in such isolation, as both the change in population being tested and their health seeking behaviour may be of equal interest: e.g. a CTG may equally well identify fetal distress successfully in two women, but their outcome will differ depending on whether they presented prior to or after fetal death. The last portion of this group of questions, how stillbirth reduction can be achieved simultaneously with reduced admissions for induction of labour at the time of her first consultation for DFM, is harder to address without including too much speculations. But with the question asked, we would offer at least two possibilities for consideration. Firstly, that maternal health seeking behaviour (i.e. timeliness of presentation with DFM) would affect the urgency of interventions, so that problems would be identified at an earlier stage and thus reduce the need for immediate induction. Second, that uncertainty in clinical decision-making among health care providers would be reduced by having concrete guidelines to follow, and thus the confidence not to admit women acutely for induction “just in case” once a thorough examination had excluded urgency. The fact that we did not collect the mode of delivery in all pregnancies in the evaluation of this project is a weakness that we regret, as discussed in the paper, and thus we do not know whether the overall induction rates for women having experienced DFM has changed or not.
We hope this somewhat lengthy comment has brought clarification.
On behalf of the authors, sincerely,
J. Frederik Frøen Dept. Director, Assoc. Professor, MD, PhD Dept. of Genes and Environment Norwegian Institute of Public Health
Education/information may be necessary to decreased fetal movement especially “at term”.
24 August 2009
Dear Sir,
Although maternally perceived decreased fetal movement (DFM) sometimes precedes imminent fetal jeopardy, controversy remains as to whether universal screening, ie,, informing all pregnant women of fetal movement (FM), and identifying women with DFM followed by intervention, reduces the stillbirth rate.
We applaud Dr Tveit and her colleagues for having shown that universal screening for DFM did reduce the stillbirth rate in eastern Norway [1]. A well designed brochure [2] accompanying this article is applicable to any other institutes or countries.
Recently, we have introduced “modified count to 10” for FM, and established its reference value for low-risk Japanese [3]. As is commonly known, perceived FM decreases toward term, with the 32th week showing the highest rates of FM [3].
More recently, we encountered a clinically significant case. A 26-year-old Japanese primigravida visited us for a regular pregnancy checkup at 39+0 weeks. She informed us of DFM over the last 3 days. Unfortunately, the fetus had already died. She had recorded a “modified count to 10” chart from the 27th week. She was previously educated/informed that DFM was sign of a fetal jeopardy and that, in cases of DFM, she should consult us immediately. Unfortunately, she was also notified that FM may decrease toward term: she thought that DFM this time may be due to her being near term and preceding signs of labor onset. That was why she did not consult us immediately, and waited 3 days.
Based on Tveit et al.’s article [1] and our experience, we emphasize the following 3 points. First, it is true that FM generally decreases toward term, but this should not be over-emphasized as it may mislead pregnant women. Second, pregnant women should be encouraged to consult a care provider whenever experiencing DFM irrespective of the number of gestational weeks, even when near the expected due date. Tveit et al. [1] also touched on this issue, but did not emphasize DFM near term. Third, the point mentioned above should definitely be stated in the brochure [2]. In the brochure, it is described: “even as you approach the expected date of delivery, your baby will seldom sleep more than an hour at a time” [2]. It may be better to more clearly describe this: “although FM may decrease near term, it is a subtle and slow decrease. Whenever you subjectively feel a movement decrease, contact us immediately, even near the expected due date” Such a description may be added in the next revision of the brochure, which may markedly improve it.
References
1 Tveit JVH, Saastad E, Stray-Pederson B, Bordahl PE, Flenady V, Fretts R, et al.: Reduction of late stillbirth with the introduction of fetal movement information and guidelines –a clinical quality improvement. BMC Pregnancy and Childbirth 2009, 9:32 doi:10.1186/1471-2393-9-32 (http://www.bomedcentral.com/1471-2393/9/32 (accessed on August 20, 2009)
2 http://www.biomedcentral.com/imedia/1876556089277830/supp2.pdf (accessed on August 20, 2009)
3 Kuwata T, Matsubara S, Ohkusa T, Ohkuchi A, Izumi A, Watanabe T, et al.: Establlishing reference value for the frequency of fetal movements using modified “count to 10” method. J Obstet Gynaecol Res 2008, 34:318-23.
Shigeki Matsubara (matsushi@jichi.ac.jp)
Tomoyuki Kuwata (kuwata@jichi.ac.jp)
Correspondence: Shigeki Matsubara, MD. Professor of Department of Obstetrics and Gynecology of Jichi Medical University, 3311-1 Shimotsuke, Tochigi 329-0498, Japan
Tel.:+81-285-58-7376 Fax: +81-285-44-8505 E-mail: matsushi@jichi.ac.jp
Competing interests
None
Reduction of late stillbirths - questions to the authors
15 October 2009
I have read this paper with interest. An intervention reducing the stillbirth rate by 50% in a group of women presenting with decreased fetal movements (DFM), and 30% in the study population as a whole, is good news. However, I would like to ask the authors some questions regarding their results.
In brief, the study was a comparison of obstetric outcomes in two time periods: baseline (April – October 2005) and during an intervention period (November 2005 – March 2007). All women delivered at 14 maternity wards in Norway. The intervention consisted of:
1. An informational brochure on fetal movements (including a kick chart) was distributed to pregnant women attending the routine ultrasound scan at 18 weeks of pregnancy.
2. New guidelines for health care workers on taking care of women presenting with DFM were implemented. The guidelines recommended: “a standard clinical evaluation for all women reporting DFM, an NST, and an ultrasound scan to quantify FM, amniotic fluid volume and fetal anatomy and growth. A mother presenting with a concern of DFM was to be examined within 2 hours if absence of FM was suspected, otherwise within 12 hours”.
The authors present a reduction in stillbirth rate after 27 weeks of gestation, but no change in the proportion of women presenting with DFM (6.3% vs 6.6%, p= 0.19).
Baseline (April – October 2005):
There were 19 407 births, 50 stillbirths, and 1215 consultations for DFM.
The stillbirth rate among women with DFM was 4.2% (50/1215).
Stillbirth rate in the baseline population was 3.0/1000. Total number was not reported in the paper, but can be calculated to 58 stillbirths - 50 reported in the paper and 8 not reported in the paper, presumably among women without DFM.
During the intervention (November 2005 – March 2007):
There were 46 143 births, 73 stillbirths and 3038 consultations for DFM.
Stillbirth rate among women with DFM was 2.4% (73/3038).
Stillbirth rate in the population exposed to the intervention was 2.0/1000. From this information I can estimate 92 stillbirths - 73 reported in the paper and 19 not reported, presumably among women without DFM.
The effect on stillbirth rates was observed immediately after the intervention was implemented. Figure 2 in the paper demonstrates an abrupt downward shift in the curve from November 2005 onwards, and a significant change in stillbirth rate was observed after 7 months.
I kindly ask the authors to clarify how the change in the stillbirth rate from November 2005 onwards can be explained by the intervention?
1. Registration of FM was advocated in the third trimester (after 28 weeks). Women receiving an information brochure at 18 weeks will not come for DFM or fetal death until several months later. Hence, information to women in mid-pregnancy cannot explain the abrupt decline in stillbirth rate after study start.
2. There was no change in the proportion of pregnant women presenting with DFM (6.3% vs 6.6%, p= 0.19). It is therefore unlikely that the intervention have increased the sensitivity of counting FM as a test for fetal death risk.
3. The new guidelines included CTG and ultrasound (biophysical profile, fetal growth, Doppler). Table 1 demonstrates a marginal increase in the use of CTG (96% vs. 98%) and ultrasound (86% vs. 94%) among women presenting with DFM, but no change in the use of Doppler. These fetal surveillance tests have not reduced perinatal mortality in previous randomised controlled trials - even in high-risk pregnancies, and it is well known that the tests have poor sensitivity and specificity. It is unlikely that a small increase in the use of CTG and ultrasound can explain a reduction in the stillbirth rate by 50%.
Stillbirth rates in Norway 2005-06
The paper states: “In addition to the registrations by our study protocol, the numbers of births and stillbirths from our population were obtained from the Medical birth registry in Norway to assess overall trends in still birth, for the most updated period available: April 2005 to December 2006”.
Stillbirth rates from the Medical Birth Registry of Norway are accessible on the internet (www.mfr.no). One cannot access hospital specific rates from the ”14 delivery units in eastern Norway and the city of Bergen”. However, it is possible to analyze stillbirth rates from 8 counties in eastern Norway (Oslo, Akershus, Buskerud, Vestfold, Østfold, Telemark, Hedmark, Oppland) and Hordaland (including the city of Bergen). The deliveries as recorded in www.mfr.no in these 9 counties are, as far as I know, equivalent to the deliveries at the 14 delivery units in the study.
2005
Eastern Norway + Hordaland
N = 33 981 births. N = 70 stillbirths. Rate: 2.1/1000
Other counties
N = 22 859 births. N = 72 stillbirths. Rate: 3.1/1000
Total in Norway
N = 56 876 births. N = 142 stillbirths. Rate: 2.5/1000
2006
Eastern Norway + Hordaland
N = 35 227 births. N = 85 stillbirths. Rate: 2.4/1000
Other counties
N = 23 384 births. N = 66 stillbirths. Rate: 2.8/1000
Total in Norway
N = 58 611 births. N = 151 stillbirths. Rate: 2.6/1000
I kindly ask the authors to clarify why the study reports a 30% reduction in the population, when data from the Medical Birth Registry of Norway shows no reduction in the stillbirth rate in the counties included in the study during 2005-06. It would be helpful if the authors report the number of stillbirths in 2005 and 2006 among women:
1. with DFM in the 14 delivery units
2. without DFM in the 14 delivery units
3. who did not deliver in the 14 delivery units, but in the 9 counties
Reduced rates of preterm births and inductions of labour after intervention
The authors report a borderline statistically significant reduction in preterm births (12% vs. 10%, p = 0.1) and a statistically significant reduction in admissions for induction of labour (7.0% vs. 4.9%, p = 0.03) after a multivariate analysis. In addition, they report a small, non-significant reduction in admissions for emergency section (1.8% vs. 1.2%, ns.).
An effective intervention that reduces the stillbirth rate by 30-50% is likely to be accompanied by an increase in the number of induced deliveries, the numbers of preterm births and Cesarean sections. This study reports a statistical significant reduction of induced deliveries and preterm births during the intervention. I kindly ask the authors to clarify how a reduced stillbirth rate can be accomplished by a reduction in preterm births and in inductions of labour.
Kind regards
Kjell Å. Salvesen
Professor
National Center for Fetal Medicine
Trondheim University Hospital
N 7006 Trondheim
Competing interests
None declared
Reduction of late stillbirth - reply from the authors
29 October 2009
Dear Sir,
Salvesen’s comment to our study has just come to our attention, and we appreciate the opportunity to clarify. The comments have questions regarding three aspects of our study:
1. How come the effect is not reflected in publicly available data?
2. How come mortality seems lower already in the first month of intervention?
3. How come the intervention seems to lower mortality without increasing maternal concern and/or interventions?
1) To the first question, on why the effects cannot be seen on the web site of the Medical Birth Registry of Norway (MBRN), we regret that Salvesen has had the misfortune of using incorrect 1) Inclusion criteria, 2) Catchment criteria, 3) Catchment area, 4) Time period, 5) Intervention period, and 6) Gestational age.
1. The inclusion of pregnancies differs as our study only includes singleton pregnancies, while the MBRN web site Salvesen refers to does not stratify by pluralities and different mortality rates.
2. The catchment criteria differ. Salvesen interprets data from the MBRN web site as births in nine counties, but as specified in the first column of the tables the MBRN web site tables are not based on place of delivery, neither by county nor by institution, but on the legally registered maternal residential address. Norwegian women may attend any hospital of their choice free of charge, and frequently choose to receive their care for pregnancy and childbirth in another county if this hospital is geographically closer. The comparison with our data based on place of delivery is thus inaccurate, and further so simply because many young women live elsewhere than at her registered residential address, e.g. students.
3. The catchment area differ, as even if it was correctly based on the place of delivery, as discussed above, the simplified tables on the web site are based on counties, while our study is not. Salvesen states that ”The deliveries (…) in these 9 counties are, as far as I know, equivalent to the deliveries at the 14 delivery units in the study.”. This is unfortunate, as the MBRN web site he extracts data from lists and presents births from 22 institutions in these nine counties.
4. The time period is incorrect as these annual figures presented on the web site are incomparable with our study, and in particular the pre-intervention data are severely misrepresented. The figures presented by Salvesen add five non-included months to the seven pre-intervention months, and remove five included months from the intervention.
5. The intervention period is incorrect, adding to the inadequate time period above, as two months of data from the intervention period is moved into the pre-intervention period.
6. The gestational age may differ as many stillbirths (approximately 0.5 per 1000 births) are reported with unknown gestation at the time of reporting to the MBRN, while this may be established at a later date. In our material all stillbirths were dated after validation by the individual participating hospital.
Unfortunately, such comparisons of incompatible data sets do not add helpful information. To do the comparison, well-selected data sets from the MBRN must be used, and these are publicly available to any interested scientist through the web site BMC. While preliminary numbers for 2007 are now available on the web site, the validated data for research purposes have not yet been released, but we will, of course, take keen interest in a new dataset to compare with our own data collection. We would also recommend not to base the comparison solely on the registered gestational age, but use all singleton stillbirths reported with a gestational age > 28 weeks, or unknown gestation, in combination with the WHO criteria of 1000 grams or 35 cm body length. This would avoid loosing the data due to unknown gestation and also avoid cases of early fetal death delivered several weeks later. Our agreement with the individual hospitals is, however, that no hospital-specific data will be published.
2) Second, how the “change in the stillbirth rate from November 2005” could be explained. The presented figure is a statistical process control (SPC) chart, a time series tool of simplified statistics much used in clinical quality improvement to help understand variation and change over time. SPC provides simple rules to identify change without misinterpreting change from one measurement to the next as a “change”, but normal variation in the observed process. Thus, as noted in the legend and marked with an arrow on the chart, we make no claim that there was any change on the SPC chart before June 2006. The “change” in November 2005 might simply be normal variation, as none of the measurements are outside the control limits marked with dotted lines.
That said, we do see that the chart looks intriguing already from the very start of the intervention. Salvesen correctly notes that the information for women was distributed at approximately 18 weeks of gestation, but the description of “Registration of FM was advocated in the third trimester(…)” as the “active ingredient” of this campaign is inaccurate. Reducing this information campaign to a third trimester kick counting procedure passes the point. The purpose was to provide knowledge and information to pregnant women on how to recognize the difference between normal variation in healthy movements and worrisome signs, as well as some tools to support her decision making. Such campaigns typically create substantial attention initially and then loose the strength of novelty (in accordance with the SPC chart of this project). The launch of the intervention included not only handing out information to the first few women, but all midwives, community midwives and nurses, general practitioners, obstetricians, delivery units, antenatal care clinics, etc, in our catchment area were contacted with information in writing and/or with visits and meetings. There were news stories in local newspapers and on radio programs, pregnancy-related web sites linked to our project web site and sparked discussions in their forums where pregnant women at all gestations participated. If at any time there was a historically high awareness towards fetal activity in the catchment area, it would have been during the early phase of our intervention.
Identifying the exact portion of such an information campaign that “explains” the effects seen is obviously difficult. We have, however, submitted a secondary paper to this study that deals specifically with how the campaign affected maternal attitudes and health seeking behaviour in different high and low risk segments of our population, and hope this will bring additional clarification when published.
Also, as discussed more below, the information campaign was only one of two integral pieces of this quality improvement project, as the new guidelines for professional management were also implemented at the same time.
3) The third group of questions (2 & 3 and the final two sections) ask us to discuss further how an “information and management bundle” like our intervention can reduce adverse outcomes with apparently no downsides.
The first notion that no increase of women presenting with DFM is equivalent to no improvement in “sensitivity of counting FM as a test for fetal death risk”, is mistaken in two aspects: Firstly, it is a misconception of the term sensitivity. Increasing the total number of test positives has no implications for the sensitivity of the test. This will depend on whether the test was positive in a sick or healthy pregnancy, as the sensitivity is defined as the proportion of sick pregnancies in the population that are captured by the test. Secondly, we have clearly had no wish to increase the proportion of women reporting DFM, and the inaccurate notion noted above, that the intervention was “kick counting”, plays a role also for this question. The purpose of the information provided both to pregnant women and their care takers aimed to increase the number of women reporting DFM when they were rightly concerned for DFM, while reducing the number of reports of DFM from women who were probably unneededly concerned for DFM. This is what would change sensitivity, and we do not claim that “kick counting” is what provides improved sensitivity.
In terms of whether a small increase of fetal surveillance with CTG, ultrasound and Doppler (the latter not a part of our intervention) per se would explain all the reduction in stillbirths, we clearly agree with Salvesen that this would be highly unlikely. We also agree that most tests “have poor sensitivity and specificity”, but Salvesen’s generalized claim that none are effective in high risk pregnancies seems incompatible with evidence. Yet, the general merit of fetal surveillance is beyond this discussion and the question is rather how to improve its predictive value. As discussed above, what drives such change (given unchanged technical abilities of the test) is the proportion of sick pregnancies among those being tested. We would therefore suggest not to consider the correlation between outcomes and test prevalence in such isolation, as both the change in population being tested and their health seeking behaviour may be of equal interest: e.g. a CTG may equally well identify fetal distress successfully in two women, but their outcome will differ depending on whether they presented prior to or after fetal death.
The last portion of this group of questions, how stillbirth reduction can be achieved simultaneously with reduced admissions for induction of labour at the time of her first consultation for DFM, is harder to address without including too much speculations. But with the question asked, we would offer at least two possibilities for consideration. Firstly, that maternal health seeking behaviour (i.e. timeliness of presentation with DFM) would affect the urgency of interventions, so that problems would be identified at an earlier stage and thus reduce the need for immediate induction. Second, that uncertainty in clinical decision-making among health care providers would be reduced by having concrete guidelines to follow, and thus the confidence not to admit women acutely for induction “just in case” once a thorough examination had excluded urgency. The fact that we did not collect the mode of delivery in all pregnancies in the evaluation of this project is a weakness that we regret, as discussed in the paper, and thus we do not know whether the overall induction rates for women having experienced DFM has changed or not.
We hope this somewhat lengthy comment has brought clarification.
On behalf of the authors, sincerely,
J. Frederik Frøen
Dept. Director, Assoc. Professor, MD, PhD
Dept. of Genes and Environment
Norwegian Institute of Public Health
Competing interests
None declared