Maternal mortality at the AKH compared to equivalent European institutions

In addition to mortality data from the Obstetric Clinics at the AKH between 1784 and 1858, Figures 1 and 2 include data for the Dublin Maternity Hospital for the same years, used as ‘control’ data since mortality rates did not differ significantly during this time. Obstetricians in Dublin Maternity Hospital did not perform autopsies before entering the labour wards. In addition they washed their hands and had some understanding that disease could be contagious, although they considered other possibilities as well. Data from the obstetrical clinics in Budapest, managed by Semmelweis between 1851 and 1859, were also analysed. Earlier data from Hungarian clinics were destroyed in the abortive Hungarian uprising, crushed in early 1849, against the absolutist rule of the Hapsburg monarchy [Reference Waller5].

1784–1822: Baseline data

The mortality in the maternity hospitals in Vienna and Dublin was the same for the years 1784–1822 (Fig. 1). Furthermore, the maternal mortality exceeded 2% in only four of those 39 years in Dublin and in only seven of those years in Vienna (Fig. 1).

1823–1847: The effect of the replacement of Professor Lucas Boër by Professor Johann Klein and the introduction of teaching in Pathological Anatomy

In January 1823 (Figs 1 and 2, line A) Professor Lucas Boër, who had trained in England and subscribed to the theory that puerperal sepsis was a contagious condition that could be spread from infected patients, and for which transmission could be reduced by general hygienic principles, was ousted as Director of the AKH on ‘the pretext of insubordination’ [Reference Slaughter9, p. 17]. He refused to use cadavers in the teaching of obstetrics to Viennese students. Instead he instructed students using a ‘phantom’ or mechanical mother. His refusal was seen as resistance to the developing trend to increase the ‘anatomical’ basis of medical teaching in Vienna.

Boër was replaced by his former assistant Professor Johann Klein, who opposed the contagion theory. In the same year the post-mortem dissection of the deceased and the formal study of their diseased organs, Pathological Anatomy, was introduced into the teaching curriculum for (male) Viennese medical students. As the first Professor of Pathological Anatomy at the University of Vienna between 1844 and 1875 Karl (von) Rokitansky would become one of the outstanding pathological anatomists of the 19th century and would perform in excess of 30000 autopsies during this period [Reference Semmelweis8, p. 95].

Therefore, beginning in 1823, when Klein became professor, male doctors and male medical students at the University of Vienna were exposed for the first time to post-mortem dissections, including those mothers who had died from puerperal sepsis. In addition male students of obstetrics practiced vaginal examinations on recently deceased mothers. They would also perform those same examinations on healthy mothers in labour in the AKH, often going directly from the morgue to the maternity ward.

Mortality increased significantly in the AKH compared to Dublin Maternity Hospital following the appointment of Johann Klein as Professor of Obstetrics in 1823 (Figs 1 and 2); and rose still further and remained higher than in Dublin until 1846.

1833–1840: The effect of the division of obstetrical services at the AKH into Clinics 1 and 2, both staffed by (male) medical students and (female) midwives

After January 1833 (Figs 1 and 2, line B), an increasing workload at the AKH required that the work be shared between two clinics, subsequently known as Clinic 1 and Clinic 2, which rotated the admission of expectant mothers every 24 hours. The maternal mortality for both clinics combined for the years 1833–1840 was not different from the rate between 1823 and 1832 but was significantly higher than the rate prior to the appointment of Professor Klein in 1823 (Fig. 1). Mortality was also higher in both clinics than in the Dublin Maternity Hospital during the same period (Fig. 1). However, mortality in Clinic 1 was not different from Clinic 2 during those years (1833–1840) (Fig. 2).

1840–1846: The effect of staffing Clinic 1 exclusively with male medical students and male obstetricians and Clinic 2 with female midwives

By Imperial Decree of 19 October 1840 (Fig. 2, line C), the staffing of Clinics 1 and 2 was arranged according to gender (and hence medical speciality) since gender determined the medical speciality to which students could aspire at the University of Vienna. As a result, female midwives and female students in midwifery were assigned to Clinic 2 whereas male obstetricians and male medical students were allocated to Clinic 1. The mortality in Clinic 1 was significantly higher than in Clinic 2 for the years 1841–1846 (Fig. 2). Admission to Clinics 1 and 2 was assigned by rotation and changed every 24 hours, except on weekends when all admissions were to Clinic 1. In his book Nuland [Reference Nuland6, pp. 20–21] describes the following fictional interaction between a pregnant mother and a medical student controlling admissions to Clinic 1: ‘Please pay attention, young miss! Patients are divided equally and alternately by the days on which they come to us. From Friday afternoon until Sunday afternoon we admit to the First Division. You must go to the doctors!’ The additional point of course was that the higher death rate in Clinic 1 was apparently an open secret amongst the women in Vienna; ‘women particularly dreaded the first clinic and tried desperately to avoid being admitted there’ [Reference Carter and Codell2, p. 17]. Whilst this might have produced bias, it seems improbable that women at lower risk of puerperal sepsis (for whatever reason) could have willingly delayed the onset of labour for up to 24 hours (weekdays) or 48 hours (weekends).

Thus, as described by Semmelweis [Reference Carter and Codell2, p. 64], admissions to the first clinic began at 4 o'clock on Mondays and ended at 4 o'clock on Tuesdays, at which time admission to Clinic 2 began. Then at 4 o'clock on Wednesdays admission to Clinic 2 halted, and resumed in Clinic 1. This alternation of admission to each of the clinics continued on a 24-hour basis except between 4 o'clock on Friday afternoons and 4 o'clock on Sundays when admission was exclusively to Clinic 1. From Sunday afternoon to Monday at 4 o'clock admissions to Clinic 2 resumed. Thus, Clinic 1 received mothers 4 days a week and Clinic 2, 3 days a week. Mortality in both clinics was also significantly higher than in Vienna prior to 1823 and in Dublin in 1841–1846. This indicates that a unique factor (or factors) was present in Clinic 1 explaining the higher maternal mortality than in Clinic 2 whereas additional common factor(s) present in both clinics, and introduced only after 1823, explain their higher mortalities than in the Dublin Maternity Hospital.

1846–May 1847: The appointment of Semmelweis as Assistant to Professor Klein in Clinic 1 and the effect of the death of Professor Jakob Kolletschka on Semmelweis's evolving hypothesis

On 1 August 1844 Semmelweis began his appointment as aspirant Assistant to Professor Klein in Clinic 1 at the AKH. On 27 February 1846 (Figs 2 and 3, line D) Semmelweis was appointed provisional Assistant in Clinic 1. Then on 1 July 1846 Semmelweis was promoted to Assistant (Fig. 3, line E) to replace Dr Franz Breit whose 2-year contract had expired. Figure 3 (line F) indicates the date (20 October 1846) on which Semmelweis's position was withdrawn in favour of Dr Breit who was offered a 2-year extension to his previous contract as Assistant. As a result, Semmelweis spent the period studying obstetrical practices in England and Ireland. Semmelweis recorded that Breit ‘seldom visited the morgue in the months of December 1846 and January, February, and March 1847. The Austrian students … followed his example’ [Reference Semmelweis8, p. 102]. It will become apparent why the absence of the less conscientious Breit and his students from the morgue led to a dramatic fall in maternal mortality (Fig. 3, line 1) during those months.

Figure 3 (line G) indicates the death of Semmelweis's friend Jakob Kolletschka, the Professor of Forensic Medicine, from disseminated sepsis in March 1847. Kolletschka's death was caused by a scalpel wound received whilst conducting an autopsy on a mother who had recently died from puerperal sepsis. The autopsy findings on Kolletschka provided the moment of discovery after which Semmelweis understood how puerperal sepsis was transmitted: ‘Day and night I was haunted by the image of Kolletschka's disease and was forced to recognise … that the disease from which Kolletschka died was identical to that from which so many maternity patients died’ [Reference Semmelweis8, p. 19].

To arrive at this conclusion, Semmelweis had to discount at least two entrenched dogmas of the time; that puerperal sepsis was a disease affecting exclusively women and that this disease, like all others, had multiple (atmospheric-cosmic-terrestrial) or ‘telluric’ causes [Reference Semmelweis8, p. 65].

Discarding these dogmas, Semmelweis concluded that: ‘I was forced to admit that if [Kolletschka's] disease was identical to the disease that killed so many maternity patients, then it must have originated from the same causes that brought it on in Kolletschka’ [Reference Semmelweis8, p. 19]. Next he realized that as the scalpel had introduced ‘decaying matter’ into the blood of Kolletschka, so the hands of the obstetricians and students, soiled by their contact with the tissues of the recently diseased, were passing the same material to the birth canals of the mothers in labour. Semmelweis identified his own deadly contribution: ‘only God knows the number of patients who went prematurely to their graves because of me. I have examined corpses to an extent equalled by few other obstetricians’ [Reference Semmelweis8, p. 98]. Thus Semmelweis ‘identified the contaminated hands of the examining physicians as the source of the decaying matter that spread the same disease [that had killed Kolletschka] among the patients’ [Reference Semmelweis8, p. 20].

Hence, Semmelweis correctly deduced that the cause of the excess mortality from puerperal sepsis in Clinic 1 was the transfer of some agent, present in and on the bodies of the deceased mothers, from the autopsy room to the vaginas of the mothers in labour, on the hands of the obstetricians and students assigned to Clinic 1, thereby sustaining the infectious cycle since labelled ‘The Doctor's Plague’ [Reference Nuland6]. Since Semmelweis was not at that time working in the AKH, he was temporarily powerless either to test his novel hypothesis or to transfer ‘cadaverous particles’ from the morgue to the maternity wards.

Dr Franz Breit was appointed to the Chair of Obstetrics at Tubingen University on 20 March 1847 (Fig. 3, line G), allowing Semmelweis to be reappointed in his stead as Assistant in Clinic 1 at the AKH. The dramatic changes in mortality when Semmelweis was first replaced and then reinstated in Clinic 1 can now be explained. The steeply falling mortality in the period when Semmelweis was replaced by Breit in Clinic 1, November 1846 to March 1847 (Fig. 3, line 1, combined data for Clinics 1 and 2) can be explained by Breit's disinterest in performing autopsies. As a result, neither he nor his students carried ‘cadaverous particles’ to the mothers in Clinic 1 during those months. Conversely the immediate increase in mortality after Semmelweis’ reinstatement as Aspirant in Clinic 1 in March 1847 (Fig. 3, line 2, combined data for Clinics 1 and 2) was due to the resumption of his autopsy duties, after which he returned to Clinic 1 to fulfil his obstetrical duties including examining, with soiled hands, mothers in labour.

May 1847–1849: The effect of the introduction by Semmelweis of handwashing with chlorine in Clinic 1

Realizing that his proposed cycle of infection would be broken if all obstetricians and medical students in Clinic 1 washed their hands with chlorine (chlorina liquida) before examining the mothers in labour, Semmelweis introduced chlorine handwashing into Clinic 1 in late May 1847 (Figs 2 and 3, line H). Semmelweis also introduced a policy of public shaming in which the names of the midwife and student assigned to each mother in labour were displayed above her bed ‘for all to see’ [Reference Waller5, p. 140]. Should the mother die from puerperal sepsis those responsible for her death were immediately identifiable.

Following the introduction of the chlorine washings the mortality in Clinic 1 fell sharply, so that between 1847 and 1849 mortality rates in Clinics 1 and 2 were no longer different (Fig. 2). Furthermore, mortality in Clinic 1 in those years was not different from the mortality in the AKH in the years before the appointment of Professor Klein and the introduction of an ‘anatomically based’ education in 1823 (Fig. 2). However, mortality in Clinic 1 was higher than in the Dublin Maternity Hospital during the same years (Fig. 2).

Semmelweis concluded that

Further support for Semmelweis' theory arose in March 1848 (Fig. 3, line K). As widespread civil uprising against the rule of the Hapsburgs reached Vienna in that month [Reference Waller5], clinical teaching of medical students at the AKH was suspended and Semmelweis himself was ‘unusually absent from work’ [Reference Nuland6, p. 122]. As a result, not one of 276 mothers who gave birth in either clinic during that month developed puerperal sepsis [Reference Semmelweis7, p. 536]. A similar result was achieved in August the same year. Never before had this occurred.

But other clinical experiences had taught Semmelweis that his initial explanation for the spread of puerperal sepsis was too restrictive. Since he had already concluded that there must be only one singular cause for puerperal sepsis, Semmelweis assumed there had to be a logical explanation for the increased mortality that he observed in Clinic 1 in October and November 1847 and which caused the increased mortality in the combined data for Clinics 1 and 2 (Fig. 3, line 4). The increased mortality in October (Fig. 3, line I) was traced to the presence of a patient in Clinic 1 with a discharging medullary carcinoma of the uterus occupying the bed at which the ward rounds always began. Since doctors washed their hands with chlorina liquida only before they examined the first patient on the ward round and then used only soap and water between the internal examination of subsequent patients, Semmelweis noted that:

In November 1847 (Fig. 3, line J) a patient with a discharging ‘carious’ knee was admitted to the ward. As a result, according to Semmelweis:

Unexplained by Semmelweis's theory was why these two mothers had not themselves succumbed to puerperal sepsis.

As a result of these additional experiences, Semmelweis concluded that:

To explain why this explanation was not immediately embraced, Waller [Reference Waller5] points out that, at least in Britain and France, morgues were seldom located in the maternity hospitals. Thus Semmelweis's theory appeared to explain a special case, specific only to the AKH. In particular, Semmelweis's theory could not explain why mothers in Clinic 2 continued to die, or why puerperal sepsis also occurred in mothers delivering outside hospital. The germ theory of infection explained the universal applicability of Semmelweis's theory [Reference Waller5].

After 1850: The effect of the termination of Semmelweis's contract as Assistant to Professor Klein in Clinic 1 and his replacement by Dr Carl Braun

Figure 3 (line L) records when Semmelweis's contract in Clinic 1 was terminated (March 1849). He was replaced by Dr Carl Braun who would later succeed Klein as Professor of Obstetrics at the University of Vienna. Braun did not accept Semmelweis's theory. As a result, once Semmelweis had left Vienna, Braun did not enforce the prophylactic measures introduced between 1847 and 1849.

On 15 October 1850 Semmelweis left for Budapest (Figs 2 and 3, line M) where he was appointed honorary Senior Physician in the Obstetrics Division at St Rochus Hospital in Budapest and in July 1855 as Professor of Theoretical and Practical Midwifery at the University of Pest.

Waller [Reference Waller5] offers the novel explanation for why Semmelweis left Vienna even though a group of outstanding young professors at the University of Vienna supported his revolutionary ideas. He argues that Semmelweis actively identified with the republican forces wishing to overthrow the Hapsburg monarchy whereas Klein was an ardent monarchist. Klein saw Semmelweis's theory as a revolutionary threat to that hierarchy and so it had to be suppressed. According to Waller, Semmelweis had no option but to leave Vienna ‘a belated casualty of the year of revolutions’ [Reference Waller5, p. 148].

Following the appointment of Carl Braun, mortality in both clinics at the AKH began to rise (Figs 2 and 3, line M). Thus mortality was significantly greater in both clinics than prior to 1823, although not statistically higher than the mortality between 1847 and 1849 when chlorine washing was in vogue (Fig. 2). Later Semmelweis would calculate that during Braun's 12-year tenure an excess of 1035 deaths occurred because the practice of chlorine washings was not promoted [Reference Semmelweis7, p. 856].

1850–1858: Maternal mortality rates in the Budapest Maternity Hospital directed by Semmelweis

In contrast, in his new position in the maternity hospitals in Budapest, Semmelweis initially maintained a low mortality in Budapest until 1857 and 1858. Semmelweis explained the greater mortality in those 2 years on soiled bed linen that was not changed between deliveries nor properly sterilized because the hospital outsourced the cleaning of linen to the cheapest contractor [Reference Semmelweis7, pp. 424–427]. As a result, the mortality in the Budapest hospitals during the years of Semmelweis's direction (1850–1858) (range 0·39–4·01) were not significantly different from Clinic 2 of the AKH (range 1·06–6·18) during the same time (P=0·09). However, for 8 out of the 9 years Clinic 1 of the AKH had a higher mortality rate (range 1·07–9·10) than the Budapest hospitals (range 0·39–4·01) (P=0·02) indicating a substantial saving of maternal lives in Budapest compared to Vienna. Interestingly the mortality rate in Clinic 1 remained lower than in Clinic 2 between 1849 and 1852 even though Semmelweis was no longer present. This suggests some residual effect of his teachings.