The tip of the iceberg: postpartum suicidality in Israel

Reported here are data from various sources that offer an estimation of the rate of suicidal ideation among Israeli women in the postpartum period (Table 1) [30‐35]. The most current and broad data, albeit as yet unpublished, were communicated from the MOH Tipat Chalav database (Rubin, L., personal communication, 2017) and Maccabi Health Services database (Fish, R., personal communication, 2017). All data is based on responses to the Edinburgh Postnatal Depression Scale [26, 36] (EPDS), the most widely used instrument in research dealing with perinatal mental health. The EPDS is a 10-item screening instrument that includes a question regarding thoughts of self-harm (Question 10) that states: “In the past week the thought of harming myself has occurred to me: quite often/sometimes/hardly ever/never.” Any response other than ‘never’ is considered suicidal ideation, and guidelines instruct immediate consultation in the event of this response [25]. After its successful use in Israel in a pilot project in MOH Mother-Child Healthcare clinics [30], its implementation for universal screening of depressive symptoms during pregnancy and within the first postpartum months was gradually expanded, and since 2013 has been mandatory.

Data on non-fatal suicide attempts were obtained from the National Hospital Emergency Department [ED] database, which is maintained by the MOH. This includes all ED admissions, with demographics, reason for admission, diagnosis, date and time of admission and discharge, and ED discharge destination. Individuals’ identity numbers are encrypted to protect patient privacy, but allow matching of records belonging to the same individual. The data were cross-checked by encrypted identity number with the National Birth Registry, to identify women who were admitted to ED for a suicide attempt within one year of delivery.

The present analysis included individuals admitted to the ED in all general hospitals in Israel during 2006–2015 whose reason for admission was classified as attempted suicide and/or had an ED diagnosis of suicide (ICD-9 codes E950-E959), and were aged 18 to 44 years-of-age. Psychiatric hospitals and those in East Jerusalem are not included in this database.

Suicide attempts were characterized by age group (18–24; 25–34; and 35–44) and ethnic group. As of 2016 the population of Israel was comprised of 74.8% Jews, 20.8% Arabs, and a small percentage (4.4%) of “Others” [non-Jewish Israelis who are not Arabs, many of whom are immigrants or relatives of immigrants from the Former Soviet Union (FSU)]. For this analysis, the data presented for Jews include “Jews and Others.” The group of Jews was further divided into four sub-groups: Israeli-born or veteran immigrants (before 1990), and recent immigrants (since 1990) from either the FSU, Ethiopia, or from other countries. The postpartum period was defined as one year following childbirth. The control group included all other suicide attempts recorded in the ED database for women in this age group.

The population denominator used to calculate rates in the postpartum group was taken as the mid-year population of women after birth for each year and group, estimated from the National Birth Registry as those within a year of birth on June 30th of each year. The control group populations were calculated as the difference between the postpartum population and mid-year total population estimates provided by the Israeli Central Bureau of Statistics (CBS) for each age and ethnic/immigrant group.

Suicide attempt rates and rate ratios with 95% confidence intervals (CI) were calculated for the postpartum and control group by year, ethnic group, age, and immigrant status group for the study period. To assess the effect of all characteristics together, a multivariate logistic model was built predicting suicide attempt by postpartum status, age group and ethnic/immigrant group.

Data on completed suicides of women aged 18–44 were taken from the national database of causes of death, maintained by the CBS, based on death certificates. Since 1998, the CBS has coded the underlying cause of death according to the International Classification of Diseases Version 10 [ICD-10]. This data was cross-checked with the National Birth Registry to identify cases of maternal death within one year of delivery, for the period 2006–2015.

Suicide and suicide attempt data were linked and analyzed with SAS 9.4 software (Cary, NC: SAS Institute Inc.).

The study was approved by the IRB committee of the Israel Ministry of Health (MOH 029–2017).

Several studies in Israel have reported response rates to Question 10 on the EPDS during the postpartum period (Table 1). The rate of positive responses has ranged from 8.7% in an early study in a low socioeconomic area with a large proportion of new immigrants, to 1% or less in more recent studies of the general Israeli population. Higher rates, of 2–3%, have been reported in studies of Israeli Arab women. A study conducted by the Clalit Health Fund [36] (not shown), enabled analysis of the data by geographical region and also by type of community (urban, rural, etc.). In that cohort the rates of suicidal ideation ranged from 0.8 to 3.3% among the regions and from 0.8 to 5.7% among the different types of community.

During 2006–2015, 20,259 cases of suicide attempts by women aged 18–44 were recorded attending the EDs of general hospitals in Israel. Of these attempts, 565 (2.8%) were by women in the postpartum period. Fig. 1 presents the suicide attempt rate for postpartum compared to non-postpartum women by year. The suicide attempt rate for non-postpartum women was three- to five-times that of postpartum women. The rate among non-postpartum women, 144 per 100,000 population in 2006, was relatively stable until 2009 and then increased, reaching 164 in 2015. For postpartum women, the rate varied between a peak of 42 per 100,000 in 2011 and 2015, with the lowest rate of 29 in 2009.

The suicide attempt rate for the total period 2006–2015 was 36 per 100,000 population for postpartum women, and for non-postpartum women it was 153, i.e., 4.3 times greater. Suicide attempt rates for both postpartum and non-postpartum women decreased with age, and the rate ratio non-postpartum/postpartum increased (Table 2). Rates were lower for Jews in both groups compared to Arabs, and the rate ratio non-postpartum/postpartum was higher for Jews, 5.7 (95% CI 5.1–6.4) compared to 2.7 (95% CI 2.4–3.1) for Arabs.

Among Jewish postpartum women, immigrants from the FSU and Ethiopia had the highest suicide attempt rates, while for non-postpartum women FSU immigrants had higher rates, but Ethiopians did not. Hence the rate ratio non-postpartum/postpartum was lowest for Ethiopians, 1.5 (95% CI 0.8–2.7) compared to 2.9 (95% CI 2.3–3.8) for FSU immigrants and 6.6 (95% CI 5.7–7.6) for Israeli-born or veteran immigrants, respectively.

Comparison of the suicide attempt age distribution for Jews with that of Arabs in the postpartum and non-postpartum groups (Fig. 2), shows that for non-postpartum women it was very similar for both the Jewish and Arab women, with about half of the attempts in the 18–24 age group and about 30% in the 25–34 age group. However among postpartum women, the distribution differed; while for Arab women almost half of the attempts (48%) were in the 18–24 age group and only 7% in the 35–44 age group, the corresponding proportions for Jewish women were 29 and 18%.

Multivariate logistic regression analysis predicting the risk for suicide attempt (Table 3) showed similar results to those reported above. Non-postpartum women had a fourfold higher risk of suicide attempt than did those within a year postpartum; when controlling for age and ethnic/immigrant group, the adjusted odds ratio (AOR) was 4.08 (95% CI 3.75–4.44). Arabs had a significantly higher suicide attempt risk compared to Israeli born/veteran immigrants (AOR = 1.37, 95% CI 1.32–1.41), even when controlling for age and postpartum status, as did immigrants from the FSU (AOR = 1.39, 95% CI 1.33–1.44). However, Ethiopian immigrants did not have a significantly different risk when controlling for the other factors. The risk of suicide attempt in the youngest age group, 18–24, was over two-and-a-half times that of the oldest group, aged 35–44 (AOR = 2.67, 95% CI 2.58–2.77), while women aged 25–44 also had an elevated risk.

During the decade 2006–2015 there were 193 deaths among women aged 18–44 within the year following childbirth. Causes of death as recorded by the CBS are presented in Table 4.

About one-fifth of the deaths (38/193) were due to external causes, only seven of which were registered as intentional self-harm (suicide), comprising 3.6% of the postpartum deaths. The rate of reported suicide per 100,000 live births was 0.43. It was noted that three of these seven suicides were of Arab women.

It is likely that the true number of suicides is somewhat higher. For example, some of those with external causes classified as accidental fall, other accidents, or of undetermined intent may actually have been suicides, as well as some of those with unknown or undetermined cause of death.

From the available Israeli published reports [30‐35] and the as-yet unpublished data from the MOH and Maccabi Health Services databases, the rates of postpartum suicidal ideation, as reflected by any score other than zero on Question 10 of the EPDS screening questionnaire (thoughts-of-self-harm), are considerably lower than those reported in other countries. In a study of over 1000 women in New York, Bodnar-Deran [37] found that 6% of the participants presented with suicidal ideation during the first six months postpartum. Among 1500 pregnant women in Peru, 8.8% of the participants responded in the affirmative to Question 10 [38]. Howard [39] reported that by 18 weeks postpartum 9% of the 4150 women who completed the EPDS reported some suicidal ideation (including hardly ever); 4% reported that the thought of self-harm themselves had occurred sometimes or quite often. In that study, multivariate analysis indicated that suicidal ideation was associated with younger age, higher parity and higher levels of depressive symptoms, and endorsement of ‘yes, quite often’ to Question 10 was associated with affirming at least two clinical interview items on suicidality. In Lindahl et al.’s extensive review of this topic [19], postpartum suicidal ideation rates ranged from 4.6 to 15.4% in countries including the United States, Canada, Britain, Finland, South Africa and India.

In Israel, the only exception to the low rates is the study reporting data collected in 1995 [34], in a low socio-economic community with a large proportion of new immigrants. That rate, 8.7%, is somewhat higher than the 6.2% of suicidal ideation reported by adult females in the Israel National Health Survey, conducted in 2003–2004 [40]. The study was conducted before 2001, when the first systematic effort in Israel by the MOH was begun, raising awareness among primary care nurses and the public regarding PPD and its early detection [24, 30]. In the ensuing years, even before 2013 when the MOH mandated EPDS screening [25], attention was increasingly paid to early identification, screening and supportive intervention of maternal emotional distress by public health nurses during pregnancy and in the first two months postpartum [31, 35]. Interestingly, a U.S. study of trends in PPD symptoms [41] found an overall decline from 14.8% in 2004 to 9.8% in 2012 among thirteen states that had data over this period. It is possible that the increasing awareness and intervention, in some cases beginning during pregnancy, has contributed to the lower rates of postpartum suicidal ideation reported in more recent years, as seen in Table 1.

The rate of suicide attempts was considerably lower among postpartum women compared to non-postpartum women for all years and in all groups considered in this analysis. This was also the conclusion of Lindahl et al.’s review of 27 studies [19]. The relatively stable rate among Israeli postpartum women between 2006 and 2015, with some years having lower rates, may also reflect the impact of increasing awareness due to the universal screening program, particularly since the rate among non-postpartum women increased considerably over this period.

The overall rate of postpartum suicide attempts in Israel between 2006 and 2015 was 35.8 per 100,000 population, lower than that reported by Schiff et al. [11] of 43.9 per 100,000 live births in Washington State. The difference is actually greater, since Schiff et al. reported only hospitalized suicide attempts, while this study included all ED-admission attempts, of which only 43% were hospitalized in the postpartum period. In Taiwan, Weng et al. [13] reported even lower rates of 9.9 per 100,000 live births, but they also appear to have identified only women admitted to hospital following ED admission for serious suicide attempts, since they found only 139 attempts in 2002–2012, very few compared to the 95 completed suicides.

Some groups were found in this study to be at a higher risk than others. For example, risk was highest for mothers in the youngest age group, similar to the results of Schiff et al.’s [11], and Gressier et al.’s [5] finding that among women hospitalized in psychiatric Mother-Baby Units, younger age was a risk factor for postpartum suicide attempt. Arab postpartum women were at a higher risk, with lower rate ratios compared to non-postpartum women. Contributing to this was the higher proportion of young suicide attempters among the postpartum Arab women compared to Jewish women. One factor might be the inequalities in health service utilization between the Jewish and Arab sectors, with less utilization of specialist and mental health services among Arab women [42, 43]. This also may reflect the younger median age at which Arab women give birth, which has remained stable over the past decade, compared to that of Jewish women, which has been rising [44]. Even the age-adjusted odds ratio showed a 37% higher suicide attempt risk for Arab women compared to Jewish Israeli-born/veteran immigrants. The age-adjusted risk was also higher for postpartum FSU immigrants, similar to their higher risk for suicide and suicide attempts reported in the general population [45]. Indeed, the higher suicide attempt rates in both groups reflect the heightened stress experienced by these groups; Arab women as members of a disadvantaged minority as well as women’s subordinate position in their traditional, patriarchal community, and FSU immigrants with the stresses of immigration and the high rate of single mothers in this group [46, 47].

In the present study postpartum suicide attempt rates per 100,000 population were calculated, enabling comparison with rates of the non-postpartum population; this is in contrast to other studies that only calculated the rates per 100,000 live births. However, the two rates are very similar. For example, in this study the overall suicide attempt rate was 35.8 per 100,000 population and 34.6 per 100,000 live births.

In several studies suicide has been cited as one of the leading causes of maternal death, especially, but not uniquely, among women suffering depression or with previous psychiatric history [15, 16, 19, 20]. While in Israel suicide was an important cause of death in the first postnatal year, the rate of 0.43 per 100,000 live births, or 3.6% of postpartum mortality, was considerably lower than that found elsewhere. For example, in the Canadian 15-year population-based study [14], the suicide rate among women in the postpartum year was 1.57 per 100,000 live births, which comprised 6% of postpartum mortality. Metz et al. [18] reported a rate of 4.6 per 100,000 live births in Colorado from 2004 to 2012, and Esscher et al. [16] reported 3.7 per 100,000 live births during 1980–2007 in Sweden, which amounted to18% of maternal deaths. Even assuming an underestimation of suicides in Israel of 42%, as found by Bakst et al. [27], the revised rate of 0.61 per 100,000 live births is still low compared to other studies. On the other hand, Fuhr et al.’s [48] meta-analysis of studies from 21 middle- and low-income countries found a pooled prevalence of pregnancy-related or maternal deaths attributed to suicide of 1.0%, lower than that in Israel. However, this may be an underestimate since many of the studies included reported only deaths in the first 42 days postpartum, while suicides have been shown to often occur later in the year following childbirth [21].

The postpartum suicide rate in Israel is low compared to the nationwide suicide rate. For example, the national rate in 2011–2013 for females aged 25–44, was 2.4 per 100,000 population [45], more than five times the postpartum rate. The low postpartum suicide and suicide attempt rates in Israel compared to other high-income countries are consistent with Israel’s low overall suicide rates compared to international data [45].

One factor involved may be the protective effect of religiosity, since a disproportionate number of births in Israel are to religious women, both Arab and Jewish [49‐51]. This protective effect of religiosity has been found in various societies [52‐54], and in Israel was shown by Glasser et al. [32] who reported lower rates of antenatal depression among Arab women with increasing religiosity, and by Dankner et al. [55] regarding PPD among Jewish women in Israel. Mann [56] reported that increasing religiosity antenatally was associated with lower rates of PPD, and Van Praag [57] has noted the protective effect of religion in suicide prevention. In addition to the general protective effect of religiosity, both Judaism and Islam forbid suicide. On the other hand, this may be a “two-edged sword,” since heightened stigma in religious societies regarding mental health disorders [58‐61] can lead to under-reporting of suicide and therefore lower reported, but not actual, rates. Indeed, while Russo et al. [62] found that religion was among the positive influences on their emotional well-being among Afghan women interviewed in Australia, they also noted their cultural stigma associated with mental illness, contributing to resistance to obtaining professional support.

Although in this study the suicide rate was lower among postpartum than non-postpartum women, as also reported by Lindahl et al. [19], postpartum suicides were found to comprise a higher proportion of suicidality than were non-postpartum ones. This may indicate the greater lethality of postpartum suicide attempts, an aspect supported by several reports that note the violent methods employed in postpartum suicides, such as hanging, jumping or falling [15, 16, 19, 20, 63].

Several problems arise universally when attempting to document or monitor postpartum suicide rates. While international comparisons of cause of death are based on the assumption of equivalence of coding practices and definitions, deaths during pregnancy or postpartum are divided into direct, indirect or incidental [20, 64]. Direct deaths are the result of obstetric complications (ICD-10 codes O00-O97); indirect deaths result from aggravation of a condition by pregnancy; and incidental maternal deaths occurred during pregnancy or postpartum, but were not likely caused by it. Maternal mortality statistics include direct and indirect maternal deaths. In 2012, new World Health Organization (WHO) guidelines defined postpartum suicide as a direct cause of maternal mortality thus expanding these cases and leading to increased maternal mortality rates [65].

Definitions also differ with respect to the period under consideration. Unlike the present study, which dealt with the first postpartum year, international data on maternal death often only include deaths within 42 or 90 days of childbirth, such as many in Fuhr et al.’s meta-analysis [48]. Official Israeli data on maternal mortality includes only deaths related to, or aggravated by, pregnancy and up to 42 days postpartum, while the present study reported all deaths in the postpartum year, irrespective of their association with childbirth. Other countries include all pregnancy-related deaths in reported statistics, thus including prenatal suicides [48, 63]. ICD-10 expanded the concept by defining “late maternal deaths” (> 42 days to one year postnatally) [64]. It has been noted, however, that late maternal deaths are less likely to be documented as such [20, 66]. In the UK Confidential Enquiries into Maternal Death, later deaths were found by linking death records with births in the previous year [15]. When suicides from these later deaths, not initially reported, were included as maternal deaths, suicide was the leading cause of maternal death, compared to other direct causes divided into major subgroups. The 2016 MBRRACE-UK Report [21] has similarly concluded that maternal suicide remained the leading cause of direct deaths that occurred during pregnancy or up to one year postpartum. In a review of maternal deaths in Australia, Thornton et al. [66] found that there was a nearly four times likelihood of maternal deaths from external causes in the 9 to 12 months postnatally, compared to the first three months. Thus reporting only early postpartum data would have a direct impact on maternal suicide rates reported, and may contribute to the low prevalence found by Fuhr et al. as noted above [48]. Maternal deaths related to psychiatric illness are also increasingly being included as late maternal deaths [20].

Another problem is that under-reporting of suicide in general [66, 67] and in the postpartum period specifically [20, 68, 69], is well known and may be attributed to misclassification, or stigma in some societies, as noted above. In Fuhr et al.’s review [48], the rate rose from 1.00 to 1.68% when reclassifying the leading suicide methods from injury to suicide. An in-depth investigation (primarily from police reports) of deaths in Israel with a recorded cause that could mask suicide, such as unknown cause or of undetermined intent, indicated that the suicide rate was underestimated by 42% [27]. In Israel particularly, the factors that might support under-reporting include both the heightened stigma in religious societies regarding mental health disorders [58‐61] and the Israeli culture of the “death hierarchy,” whereby fallen soldiers are at the pinnacle and suicides at the bottom [28, 29].

The strength of this study is that the suicide and suicide attempt findings are based on data for the entire Israeli population over a long period, allowing analysis of the suicide attempts by population characteristics, and comparison of postpartum rates with those of the rest of the population. There are some limitations as well. Suicidal ideation data available is primarily descriptive in nature and no in-depth analysis was presented. The relatively new MOH Tipat Chalav database could enable such analysis to gain a better understanding of the characteristics of women who express such ideation, and it is hoped that in the future this data will be available so that prevention efforts can be focused on these at-risk women. The suicide attempt data did not include East Jerusalem hospitals. This may lead to under-estimation of Arab suicide attempt rates, which may be still greater than the already higher rates compared to Jews that were reported in this paper, for both postpartum and non-postpartum women. Also, in light of the fact that some of the sub-populations were very small, interpretation of the results should be taken with caution. Data from psychiatric hospital ED’s was also not included in the database used, but since most serious suicide attempts cause physical injury, they are usually referred to general rather than psychiatric hospitals, so this should not greatly affect the results. Regarding completed postpartum suicides, due to the small absolute number recorded, a consequence of the rare nature of this event, no statistical analysis of them could be undertaken by specific socio-demographic characteristics. Access to the psychiatric history of the postpartum women was not available for this study, although in other studies a significant proportion of women who committed suicide in the postnatal period were found to have had psychiatric treatment before or during pregnancy [14, 16, 20]. Although this could not be verified in the present study, an earlier study on suicides in the general Israeli population found a greatly elevated suicide risk among those who had past psychiatric hospitalizations [70]. Thus psychiatric history is clearly a risk factor that should be assessed in contacts with postpartum women. As mentioned above, the reliability of suicide registration is also a limitation.