Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Evidence based morbidity
- Stan Goldstein (7 December 2001)
-
Re: Evidence based morbidity
- Johan Karlberg (10 December 2001)
-
Diurnal mortality gradient in Hesse, Germany, an update including caesarean deliveries
- Guenther Heller (13 December 2001)
-
Diurnal mortality gradient for planned and not planned caesarean deliveries
- Karlberg (20 December 2001)
-
Shared responsibility for neonatal outcomes
- Sybil M Barr, Martin Keszler (24 February 2002)
-
Shared responsibilities have decreased adverse neonatal outcomes in the 1990’s
- Johan PE Karlberg, Zhong-cheng Luo (2 March 2002)
|
|
|||
|
Stan Goldstein, Medical Director HCF Australia HCF, 403 George St, Sydney, Australia, 2000
Send response to journal:
|
I am grateful for another study confirming the risk of births at night. I wonder however whether there has been any attempt made to match the morbidity with the clinician staffing patterns in the hospital (or elsewhere) of birth. The coincidence of "after-hours" morbidity and shift changes should be of concern to the community, to clinicians and to managers. However there seems to be an assumption made in this and other studies that the only likely reason for an apparent increase in after-hours morbidity is the clinical staffing. This may indeed be the case. However the Swedish study, being population based, has the potential to review the staffing configurations of different sorts of hospitals, and the possible influence of differing shifts on weekends and public holidays. If indeed the increased risk relates to these shift changes then it is important that managemenbt and clinicians work together to normalise this risk. If however the correlation between night hour births and risk is higher than that of weekend or public holiday birth and risk, it seems there may be a need to determine whether there are aspects of physiology which increase risk more so than shift change. Conceivably, tired health workers perform less well. On the other hand, it appears to be an approach which doesn't meet the best standards of scientific inquiry when we allow a hypothesis to become fact without any true evidence. I am sure there would be many grateful managers throughout the world if this study could be extended to "test the hypothesis". |
|||
|
|
|||
|
Johan Karlberg, Director & Professor Clinical Trials Centre, The University of Hong Kong Hong Kong SAR, PR China
Send response to journal:
|
Dear Sir, This is a response to the quick electronic response provided by Dr S Goldstein on our paper published in the BMJ on December 8, 2001. There is no conflict of interest for any of the two responders Early Neonatal Mortality is Evidently Related to the Timing of Birth Dr S. Goldstein has reflected on our contribution in the BMJ that addressed the timing of birth and infant and early neonatal mortality in Sweden. Our study confirmed what others have shown in both Germany and Wales, i.e. that there was a higher risk for neonatal mortality if a baby was born during the night rather than during the day. The new aspect of our paper is that we could not note any change in this pattern over the 23 years of study, although the overall early mortality has decreased over the decades. Due to a large sample size of 2 million births we were also able to identify an increased mortality risk for babies born during the change of staff shifts in the morning. We admit that our study is observational in nature and that we have no support for any causative relationship. For this reason we only concluded that “The underlying causes are not clear and may be due to excess workloads, inadequate or less experienced staff on night shifts, or out of date systems for managing shift changes within hospitals.” We do not see our study as evidence for clinical practice rather as a base for discussions to identify the problem lying behind the time-of-birth mortality fluctuation. A sign in this direction is provided by various media reports in relation to our publication. One report made by the BBC (www.bbc.com) had a comment from Carol Bates of the Royal College of Midwives, UK. Who stated that she welcomed any research that looked at the impact of the patterns of care on the outcomes of pregnancy and birth, and that they will be looking at these results very closely to determine whether this research is relevant to the UK. Another comment could be found at WebMd (my.webmd.com): Douglas K. Richardson, MD, Beth Israel Deaconess Medical Center in Boston commented however that more infants delivered on the day shift are timed or planned deliveries, whereas infants delivered on the evening shift are more likely to be emergency deliveries. Dr Richardson also stated, that in order to improve quality we need a more detailed understanding of how the time-of-birth risk is translated into a higher mortality risk. These two media reports, as well as others also from Sweden, imply that the mortality fluctuation observation is of a major concern for doctors and midwives as well as for society, and that further research is needed to clarify any causative relationship. We have already continued the analysis of the Swedish birth register to see if the time-of-birth risk mortality trend pattern can be explained by factors such as induction births, length of gestation, babies born small for gestational age, Apgar score values, the location of the hospital and the week day of the delivery. The aims of this second study are to understand if babies born during the night shift are different as a group – i.e. a high risk group - in relation to babies born during the day time, and also if there are any differences in the time-of-birth risk mortality trend pattern for hospitals in large cities versus small cities. If this study shows that babies at risk for neonatal mortality are equally distributed over the 24 hours we would of course be very concerned. If however, we find that there are more high-risk babies born during the night than during the day we need to clarify why this is the case in order to implement any kind of preventive measures. J Karlberg, professor and director Z C Luo, research associate Clinical Trials Centre, University of Hong Kong, Hong Kong SAR, People's Republic of China Correspondence to: J Karlberg jpekarl@hkucc.hku.hk |
|||
|
|
|||
|
Guenther Heller, Assistant Professor Institute of Medical Sociology and Social Medicine, University of Marburg, D-35033 Marburg
Send response to journal:
|
We appreciate the work of Luo and Karlberg (1) confirming a higher early neonatal mortality for babies born at night and agree with most of their statements. However, we think that the larger absolute and relative risk differences in our previously reported results (2) should be mainly attributed to a greater random variation (due to smaller numbers of deaths), and to a different coding. We are less persuaded that they are attributable to a postulated "less accurate data quality" in our study: - A look at the reported confidence intervals reveals the rates and risk from the study of Luo and Karlberg do not differ significantly from our results (1,2). Therefore it is questionable to conclude that the process that produces these results is different. - Previously we had excluded antepartum deaths, caesarean sections, infants born before the 37th week of gestation, as well as infants with congenital malformations or hereditary metabolic diseases. The goal was to study low risk births, with very low mortality rates. The exclusion criteria were also met for unspecified and/ or minor congenital malformations. In our current approach (reported below) we used similar exclusion criteria with respect to gestational age and excluded only those infants where congenital anomalies or malformations had been given as a reason of death. In addition, we used the "up to date" dataset covering the years 1990-2000. Applying these criteria we obtained very similar death rates, absolute and relative risks as Luo and Karlberg (table1).
Table 1: Spontaneous vaginal and caesarean section
deliveries, Hesse, Germany 1990-2000
Births Deaths Rate Relative risk (95% CI) Absolute risk(95% CI)
Spontaneous Vaginal Delivery
Day 261593 38 0.15 1.56 (1.00-2.45) 0.08 (-0.00-0.17)
Night 171586 39 0.23
Caesarean Section Delivery
Day 83176 56 0.67 2.21 (1.43 -3.42) 0.82 (0.27-1.36)
Night 21468 32 1.49
To obtain a better understanding of the underlying processes we think that it is important to look at caesarean section deliveries. If the postulated decrease in the quality of medical care at night exists, we would expect a larger mortality gradient in this group. We have therefore also included caesarean section deliveries in table 1 and observed larger absolute and relative mortality gradients. This may be interpreted as a consequence of a decreased capacity for performing adequate (emergency) caesarean sections at night. Unfortunately it could be also a consequence of the fact that the babies delivered via caesarean section during the night bear more risk compared to babies born during the day. This is because low risk babies are more likely to be delivered during daytime, and higher risk babies born in the daytime are more likely to have been delivered electively rather than emergently. The scope of future research should be to disentangle this ambiguity, and begin to investigate the purported mechanisms that translate birth time into differential mortality. Again we would like to mention that due to the observational nature of such studies (1,2) the degree of evidence from such studies will be limited but experimental trials would be difficult to perform and ethically questionable (3). Günther Heller, MD, Assistant Professor, Institute of Medical Sociology and Social Medicine, Medical Centre of Methodology and Health Research, University of Marburg Douglas K Richardson, MD, MBA, Associate Professor, Department of Neonatology, Beth Israel Deaconess Medical Centre, Harvard Medical School Björn Misselwitz, MD, Public Health Researcher, Institute of Quality Assurance Hesse, Eschborn, Germany Stephan Schmidt, MD, Professor, Department of Obstetrics, Centre of Gynaecology and Obstetrics, University of Marburg P.S. To avoid further confusion, we would also like to mention that the time period we referred to in our previously reported analyses (2) was 1990-1998, not 1990-1995 as cited by Luo and Karlberg in Table 3 (1) References: (1) Luo ZG, Karlberg J. Timing of birth and infant and early neonatal mortality in Sweden 1973-95: longitudinal birth register study BMJ 2001;323:1327-31 (2) Heller G, Misselwitz B, Schmidt S. Early neonatal mortality, asphyxia related deaths, and timing of low risk births in Hesse, Germany, 1990-8: observational study. BMJ 2000;321:274-275 (3) Varma R, Vindla S, Mascarenhas L, Westgate J, Gunn A, Heller G, Misselwitz B, Schmidt S. Early neonatal mortality and timing of low risk births BMJ 2001;322:433 (Letters) |
|||
|
|
|||
|
Karlberg , Director and Professor Clinical Trials Centre, The University of Hong Kong
Send response to journal:
|
This is a response to the quick electronic response provided by Heller et al on our recent paper on diurnal variations in early neonatal mortality (1,2). There is no conflict of interest for any of the two responders We appreciate the comments from Heller et al on our recent paper on diurnal variations in early neonatal mortality. There have been a few previous studies on this issue, all are observational in nature, and all give rise to similar evidence that night-time births bear greater risk of early neonatal death (2-4). We all seem to agree that more in-depth studies are required to explore the paths and causes and thus contribute to identify the ways to improve neonatal health care. Our comment on “poor data quality” was merely based on the fact that the German early neonatal mortality rates were much below the figures for full-term live infants without congenital malformation as reported from other countries such as Sweden, Canada and US. The data quality is of course important in any research, particularly for international comparisons of studies on neonatal outcomes since there may be variations in registration practices in different countries. While we also accept the additional explanations of the previous work of Heller et al that their results may be due to greater variation from a smaller number of events, we also agree that it may also be due to the differences in quality or standards in coding or registration practices. Heller et al argue that the similarity in the relative risk estimates of the German and the Swedish studies provides support for their high data quality. This however assumes that the relative risk estimates should in reality be the same in the two settings, which very well may not be the case. Secondly, a random drop out of early neonatal deaths from the series, giving a reduction in the overall early neonatal mortality rate, can theoretically produce the same relative risk estimates. We would also like clarify that the aim of our study was to see if the Swedish birth data could confirm the observations reported by Heller et al. For this reason we used the same inclusion criteria as defined in the German study, although we included a separate analysis on the pre-term births. Caesarean births were thus excluded in our publication, but we agree with Heller et al that this group of newborns could offer additional information on time-of-birth association with early neonatal mortality. Heller et al have provided the relative risk estimates for caesarean section deliveries based on the German series showing that this group of newborn have a higher risk of early neonatal death and that there is a time-of-birth association with early neonatal mortality also for this group. We would like to provide some additional information about the caesarean section delivery group based on the Swedish data (Table 1). The data was taken from 1983 to 1995 when the register had documented if the caesarean section delivery was planned in advance, or if it was made as an emergent operation due to critical events. All babies that were born term, did not have any congenital malformations and were singletons. . First note that 97% of the planned deliveries take place during daytime, while 28% of the un-planned deliveries happen during the night time. Comparatively, night time births include a higher proportion of high risk births; the total early neonatal mortality rate was 2.34 per 1,000 for unplanned caesarean births as compared with 0.59 per 1,000 for planned caesarean births. Night time births have a higher mortality rate for both groups as compared with day time births. We conclude that night time births have an overrepresentation of high risk deliveries which may explain a portion of the time-of-birth association with early neonatal mortality. However, night time births have a greater risk of early neonatal death for both planned and unplanned births. The mortality rate figure provided in Table 1 are also for this sub-group about double in magnitude as compared with the figures presented by Heller et al from Germany (2). Table 1. Early neonatal mortality (Mort.) and survival (surv.) for caesarean section deliveries either planned or not planned in relation to the timing of birth. _________________________________________________________________________________________________________ Total Day time births (7 am – 9 pm) Night time births (9 am – 7 pm) ____________________________ ___________________ _________________________ Caesarean Early Early Early Early Early Early section neonatal neonatal Mort. neonatal neonatal Mort. neonatal neonatal Mort. delivery deaths surv. rate deaths surv. rate deaths surv. rate type n n /1000 n n /1000 n n /1000 p value* _________________________________________________________________________________________________________ Total 205 125450 1.63 124 102582 1.21 81 22663 3.56 <0.0001 Planned 30 50785 0.59 26 49238 0.53 4 1517 2.63 <0.001 Not planned 175 74665 2.34 98 53344 1.83 77 21146 3.63 <0.0001 _________________________________________________________________________________________________________ * P value (Chi square test) between the day and night time associated early neonatal mortality. J Karlberg, , MD, Ph.D., Professor and Director Z C Luo, MD, Ph.D., research associate Clinical Trials Centre, University of Hong Kong, Hong Kong SAR, People's Republic of China Correspondence to: J Karlberg jpekarl@hkucc.hku.hk References 1. Luo ZC, Karlberg J. Timing of birth and infant and early neonatal mortality in Sweden 1973-95: longitudinal birth register study BMJ 2001;323:1327-31. 2. Heller G, Richardson DK, Misselwitz B, Schmidt S. Diurnal mortality gradient in Hesse, Germany, an update including caesarean deliveries. http://bmj.com/cgi/eletters/323/7325/1327#EL3. 3. Heller G, Misselwitz B, Schmidt S. Early neonatal mortality, asphyxia related deaths, and timing of low risk births in Hesse, Germany, 1990-8: observational study. BMJ 2000;321:274-5. 4. Stewart JH, Andrews J, Cartlidge PHT. Numbers of deaths related to intrapartum asphyxia and timing of birth in all Wales perinatal survey, 1993-95. BMJ 1998; 316:657-60. |
|||
|
|
|||
|
Sybil M Barr, Neonatal Fellow Georgetown University Children's Medical Center, Reservior Road, Washington, DC, USA, 20007, Martin Keszler
Send response to journal:
|
We congratulate Drs Luo and Karlberg on their paper demonstrating a relationship between timing of birth and early neonatal mortality.1 However, we would like to add to the conclusions drawn by the authors. To suggest that increased neonatal mortality is likely to be due to one single factor - neonatal staff shift changes - is an oversimplification of what is likely to be a multi-factorial relationship. Cesarean sections were not included in the study analysis in an attempt to minimize the obstetric effect on the timing of birth. Nevertheless, obstetric management of labour – including decisions regarding augmentation of labour using drugs, and instrumental deliveries – will remain a significant contributor to the timing of birth and importantly, to the condition of the newborn. It is widely recognised that the condition of the infant at birth is a major determinant of outcome.2 Advances in obstetric care have contributed to the improved outcomes of preterm newborns. Of course, a second consideration is the effect of diurnal variation on neonatal mortality – a question as yet unanswered. The importance of this has been clearly demonstrated in other settings such as coronary artery diseases and may be of relevance to neonatal mortality.3 In summary, we feel the important issues raised by the authors – although relevant - apply as much to the obstetric healthcare services as they do to the neonatal healthcare services. Indeed, the authors observation that the largest impact was on mortality related to asphyxia strongly implicates obstetric rather than neonatal factors. 1. Luo ZC, Karlberg J. Timing of birth and infant and neonatal mortality in Sweden 1973-95: longitudinal birth register study. BMJ 2001;323:1-5. 2. Behrman RE. Newborn Infant. Nelson Textbook of Pediatrics. Ed Behrman RE. Harcourt Brace, Philadelphia, 1994. 3. Cannon CP, McCabe CH, Stone PH, Schactman M, Thompson B, Theroux P, Gibson RS, Feldman T, Kleiman NS, Tofler GH, Muller JE, Chaitman BR, Braunwald E. Circadian variation in the onset of unstable angina and non-Q -wave acute myocardial infarction (the TIMI III Registry and TIMI IIIB). Am J Cardiol 1997;79(3):253-8. |
|||
|
|
|||
|
Johan PE Karlberg, Professor & Director Clinical Trials Centre, University of Hong Kong, Hong Kong SAR, People's Republic of China, Zhong-cheng Luo
Send response to journal:
|
Shared responsibilities have decreased adverse neonatal outcomes in the 1990’s There is no conflict of interest for any of the two responders We welcome the recent interesting electronic response of Dr. Barr and Keszler on our paper that demonstrated a relationship between timing of birth and early neonatal mortality published in December 2001 in BMJ (1,2). We agree that, in addition to the most likely effect of neonatal staff shift, other reasonable speculations can also be offered. It is reasonable to attribute the particularly higher risk of asphyxia related neonatal mortality among night time births to worse obstetric care as well, and not to neonatal care. In reality, both obstetric and neonatal care is equally important for improving the survival of high-risk infants such as preterm births. As noted from a previous electronic response by Heller and our follow-up reply (3,4), the higher risk of neonatal death associated with night time birth was also observed among caesarean section deliveries (3,4). The evidence supports the need for improving both obstetric and neonatal cares during night time, and we have actually no reason to single out one of the two. We have continued to analyse the Swedish Birth Register Data during the period 1991-1990 and we hope to be able to report our results in the near future. In this analysis we have not omitted any live birth be it pre-term births, small for gestational age babies, multiple births, babies with severe or moderate congenital disease, breach births, induced or caesarean section deliveries. We have thus kept all high risk births. The preliminary results show that the early neonatal mortality rate has decreased by close to 50% for all high risk groups taken together during the past decade. However, there is still an overall diurnal variation on neonatal mortality. We hope that our analysis will be able to isolate certain high risk groups that are associated with this variation, be it related to obstetric or neonatal care or both, or hopefully none. J Karlberg, MD, Ph.D., Professor and Director Clinical Trials Centre, University of Hong Kong, Hong Kong SAR, People's Republic of China ZC Luo,MD, Ph.D., Postdoctoral Fellow Montreal Children Hospital Research Institute, McGill University, Montreal, Canada Correspondence to: J Karlberg jpekarl@hkucc.hku.hk References 1. Luo ZC, Karlberg J. Timing of birth and infant and early neonatal mortality in Sweden 1973-95: longitudinal birth register study BMJ 2001;323:1327-31. 2. Bar SM, M Keszler. Shared responsibility for neonatal outcomes. http://bmj.com/cgi/eletters/323/7325/1327#19858 3. Heller G, Richardson DK, Misselwitz B, Schmidt S. Diurnal mortality gradient in Hesse, Germany, an update including caesarean deliveries. http://bmj.com/cgi/eletters/323/7325/1327#18051 . 4. Karlberg J. Diurnal mortality gradient for planned and not planned caesarean deliveries. http://bmj.com/cgi/eletters/323/7325/1327#18116. |
|||