The Ability of ICU Admission to Detect Maternal Near Misses as Defined By The Who Near-Miss Criteria

The maternal mortality ratio (MMR) is one of the most widely reported health indices and is commonly used to assess the quality of maternal health and obstetric practice within healthcare settings. However, in many parts of the world maternal death is an increasingly rare event. Consequently the total numbers of maternal deaths used to calculate the MMR are often small, resulting in large standard errors1. This is especially true when the total population being studied is small, such as local regions, individual institutions, or for population data taken over relatively short periods of time2. This places significant restriction on the conclusions which can be drawn from MMR, and it is now being recognised that even in settings with a relatively high rate of maternal deaths, the ratio performs poorly as a health index3. It is within this context that methods to monitor rates of severe acute maternal morbidity (SAMM) has been developed.


INTRODUCTION
The maternal mortality ratio (MMR) is one of the most widely reported health indices and is commonly used to assess the quality of maternal health and obstetric practice within healthcare settings.However, in many parts of the world maternal death is an increasingly rare event.Consequently the total numbers of maternal deaths used to calculate the MMR are often small, resulting in large standard errors 1 .This is especially true when the total population being studied is small, such as local regions, individual institutions, or for population data taken over relatively short periods of time 2 .This places significant restriction on the conclusions which can be drawn from MMR, and it is now being recognised that even in settings with a relatively high rate of maternal deaths, the ratio performs poorly as a health index 3 .It is within this context that methods to monitor rates of severe acute maternal morbidity (SAMM) has been developed.particularly within the contest of individual institutions and regions.They also allow analysis of the factors which contributed to survival in a nearmiss case, rather than concentrating factors causing death alone 5 .Common MNM statistics are shown in Table 1.There has been a wide variety of definitions for MNM and SAMM 2 .Inclusion criteria in different studies have ranged from using interventions, such as emergency obstetric hysterectomy or ICU admission, to using definitions of disease entities or organ dysfunction.Defining SAMM by critical intervention, such as emergency hysterectomy or blood transfusion has also been used 6,7 .To counter this, organ dysfunction classifications have been used to highlight severe pregnancy related complications, regardless of the disease entity 8 .The most widely used of these in the context of SAMM is the Mantel criteria 8 .However, in lower resource settings the burden of recordkeeping required to fully assess organ function may be too much for poorly resourced institutions and lead to unreliable data capture 9 .Overall, the heterogeneity of definitions and inclusion criteria of SAMM have limited the power of comparisons between studies 2 .
In 2011 the WHO published guidelines on auditing MNM which aimed to offer standardised definitions for data collection methods with the aim of using near miss statistics in a similar way to MMR 10 .The working group defined near-miss as "a woman who nearly died but survived a complication that occurred during pregnancy, childbirth, or within 42 days of termination of pregnancy", and agreed a set of inclusion criteria that covered four major categories: severe maternal complications, critical interventions or intensive care unit use, life-threatening conditions and maternal vital status 10 (Table 2).

Original PaPer
www.slcog.lk/sljogSri Lanka has been exceptional in achieving WHO millennium development goals in maternal care 11 .The latest maternal mortality ratio (MMR) for Sri Lanka is 30 per 100 000 live births in year 2012 and MMR remained low compared to other South Asian countries 11,12 .Maternal mortality surveillance system adopted by the Family Health Bureau in Sri Lanka, with the help of Sri Lanka College of Obstetricians and Gynaecologists, has been instrumental for this success 11 .However, with the recent decline in the number of maternal deaths, it has become apparent that maternal mortality reviews alone may not generate enough data to help further reduce maternal mortality 4 .Consequently, attempts to use MNM as a tool to assess the quality of obstetric care could be of great value in further reducing maternal mortality and improving maternal health.Our aim was to audit all obstetric admissions to ICU and to use the data to assess the ability of ICU admission to detect true near miss, as defined by the WHO criteria.We also wanted to compare our MNMR and MMR with other studies from the region to see whether a conclusion on our performance within the region could be reliably made from comparison of data.

METHODS
This was a retrospective descriptive study of all pregnancy-related admissions to ICU between August 2010 and May 2013 in District General Hospital Ampara Sri Lanka.Patient records of pregnancy-related admissions to ICU were searched from hospital archives by three doctors independently.All women admitted to ICU for complications relating to pregnancy and puerperium were included.All women were either directly admitted to ICU or admitted to ICU following an admission to the hospital.All maternal deaths during the study period were included regardless of whether they were admitted to ICU or not.Details of maternal deaths were cross-checked with national data.The prevalence of SAMM is generally expected to be around 7.5 cases/1000 deliveries and WHO recommends a sample containing at least 20 cases of severe maternal outcomes for an audit 10 .
The maternal near-miss mortality ratio was calculated (MNM / MM).This ratio refers to the ratio between SAMM cases and maternal death.Higher ratios indicate better care 10 .Common terminology in maternal morbidity and mortality is given in Tables 1  & 2. Data on patient demographics, disease entity, investigation results, interventions and organ dysfunction as set out by the WHO near-miss criteria were entered on to a purpose built MS/Excel sheet 10 .Indication for admission to ICU was recorded in order to determine the disease entities responsible for ICU admission which were not part of the WHO disease entity criteria.Some additions to the WHO criteria were included to reflect local circumstances and help to assess the quality of management: use of uterine tamponade and magnesium sulphate were added to our intervention criteria, as these are common interventions used at our centre for managing PPH and severe pre-eclampsia respectively.Ethical approval was obtained from the Ampara District Hospital's ethical committee and each case was identified by patient record number alone, ensuring anonymity.
A literature review of all studies into obstetric near-miss in the south Asian region was performed.We used the search terms set out by a recent systematic review of all studies into SAMM 2 .We defined south Asian region as: Pakistan, India, Sri Lanka, Bangladesh, Afghanistan, Nepal, Malaysia, Indonesia and Singapore.We have excluded studies outside these regions.We excluded studies published before 2004 to ensure the data was up to date.We also excluded studies which used emergency hysterectomy alone as their inclusion criteria, as results from these studies inevitably dealt with a much narrower definition of SAMM than our own.To compare the maternal mortality and the maternal near miss rates of our study with that of the region, published data on other countries was used.A power calculation was performed for comparison with each study in order to identify the comparisons that are adequately powered (>80%) at an alpha error of 5%.

Table 1: Common maternal near miss statistics 10
Maternal near-miss (MNM) refers to a woman who nearly died but survived a complication that occurred during pregnancy, childbirth or within 42 days of termination of pregnancy.

Maternal death (MD)
is the death of a woman while pregnant or within 42 days of termination of pregnancy or its management, but not from accidental or incidental causes.
MNM ratio (MNMR) refers to the number of maternal near-miss cases per 1000 live births (MNMR = MNM/LB).Similarly to the SMOR, this indicator gives an estimation of the amount of care and resources that would be needed in an area or facility.
Maternal near-miss maternal mortality ratio (MNM:MM) refers to the number of near misses for the number of maternal deaths.

RESULTS
A total of 9,608 live births and four maternal deaths.were reported during the study period here were 124 pregnancy-related admissions to ICU (1.29% of all deliveries) including the four maternal deaths.Women who delivered elsewhere and transferred only for ICU care were excluded.Case notes of 118 ICU admissions were analysed.Two sets of notes could not be traced.The maternal mortality ratio of our institution was 42 per 100,000.One death was due to septic abortion, one from seizure disorder aggravated during pregnancy, one from heart failure complicated by combination of post-partum haemorrhage and sepsis, and one from electrolyte imbalance secondary to severe hyperemesis gravidarum.Ninety nine patients (80%) met one of the WHO near-miss criteria other than ICU admission and were classed as true SAMM. 25 patients  • Use of blood products

• Uterine dysfunction
• Uterine haemorrhage or infection leading to hysterectomy

• Maternal death
Original PaPer www.slcog.lk/sljog(20%) met no additional criteria and so were not considered true SAMM.
The maternal near-miss rate was 9.7 per 1000 and the maternal near-miss maternal mortality ratio (MNM:MM) was 23:1.The demography of the study population is shown in Table 3.The maternal near-misses according to WHO criteria are shown in Table 4.
The commonest reasons for ICU admission were pregnancyinduced hypertensive disorders; PIH, severe pre-eclampsia and eclampsia accounted to 37.7% of all ICU admissions.All patients with severe pre-eclampsia and eclampsia received magnesium sulphate bolus and infusion.Severe post-partum haemorrhage was the second leading cause of ICU admission (19.7%).All but one patient with PPH required blood transfusion and nine patients required transfusion of five or more units of blood products.Eight of these patients (33.3%) had a hysterectomy to control bleeding.Uterine condom catheter tamponade was performed in 11 of cases of PPH (45.8%), and of those patients only one went on to have a hysterectomy.Sepsis (13.1%) was the third most common reasons for admission to ICU.All the patients with sepsis received intravenous antibiotics.
A total of 44 patients (36%) met the criteria for a single organ dysfunction, and 15 patients (12%) had multiple organ dysfunction.Cardiovascular, respiratory and coagulation dysfunction were the commonest types of organ dysfunction.Of the 25 (20%) patients who did not meet other WHO inclusion criteria 11 patients were admitted for cardiac disease and cardiac monitoring, six patients for pre-eclampsia not considered severe enough to meet the WHO disease entity criteria, two patients for epilepsy, three for observation following general anaesthesia, one for thyrotoxicosis, one for severe hyperemesis gravidarum, and one for severe asthma exacerbation.
Literature review found seven studies into SAMM in the south Asian region (Table 5).No study had a MNM:MM higher than ours, while three studies had a MNMR lower than ours, although this was not statistically suitable for comparison and MMR was significantly lower in our study than when compared to the regional data [13][14][15][16][17][18][19][20] .Results from the literature review are shown in Table 5.   13 .43% had organ dysfunction, and only 36% in our study 13 .
Although using ICU admission as an independent inclusion criterion provides an easily auditable critical event, it is likely to over-diagnose the number of true SAMM 9 , even within a relatively low-resource setting such as our institution.This would not necessarily be an issue if all obstetric units had similar thresholds for admission to ICU.However, regional differences in admission policy and variable pressures on intensive care resources are likely to affect the types of cases admitted to ICU.Evidence for this may be seen in the different proportion of patients admitted to ICU which were true and false near-miss between our study and Ranatunga et al. 13 .Under-reporting is also likely to be an issue in our study, as we would have missed cases of MNM which were not admitted to ICU.Despite this, the WHO criteria provide a validated approach to classify obstetric near miss in a variety of healthcare settings and provides a way forward for standardising MNM so that it may become a powerful statistical tool in appraising the standard of obstetric care.There may be an issue with over-reporting MNM if ICU admission is used as an independent defining criterion.One way to overcome this would be to use ICU as a screening tool for MNM, with exclusion of false MNM from subsequent analysis.Alternatively, simply displaying the proportion of pregnancy-related ICU admissions which met no other near-miss criteria could serve as a useful comparative guide to the admission thresholds in the studied healthcare settings.In this context, definition based ICU admission as independent inclusion criteria for MNM could be still justified as it would provide a straightforward way to identify the more severe SAMM cases, screen them for true near miss, while indicating any great differences in admission thresholds between study populations.