Mortality in a cohort of IVF patients

Alison Venn1,2,6, Elina Hemminki3, Lyndsey Watson1, Fiona Bruinsma1 and David Healy4,5

1 Centre for the Study of Mothers' and Children's Health, La Trobe University, Carlton 3053, 2 Menzies Centre for Population Health Research, University of Tasmania, GPO Box 252–23, Hobart 7001, Australia, 3 Stakes (the National Research and Development Centre for Welfare and Health), Helsinki 00531, Finland, 4 Department of Obstetrics and Gynaecology, Monash University, Clayton 3168 and 5 Monash IVF, Richmond 3121, Australia


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: Risks associated with IVF and related assisted reproduction technologies include complications of ovarian stimulation, surgical procedures and pregnancy itself. Serious complications are uncommon but may be potentially life threatening. The aims of this study were to compare the mortality rates of women who received IVF treatment, as well as those who were referred but were not treated, with the mortality rate in the general female population, to determine the maternal mortality rate following IVF conception and to establish whether any deaths had occurred as a result of treatment complications. METHODS: Deaths were identified in a cohort of 29 700 Australian IVF patients by record-linkage with the National Death Index and a cancer registry. RESULTS: The all-cause mortality rates in IVF patients (treated and untreated) were significantly lower than in the general female population of the same age. In treated women, 72 deaths were observed and 125 deaths were expected giving an age-standardized mortality ratio of 0.58 (95% confidence interval 0.48–0.69). Two maternal deaths were identified in the 42 days of the puerperium. Complications of ovarian hyperstimulation syndrome could not be directly related to any of the deaths identified in this cohort. CONCLUSIONS: As well as providing some reassurance about the safety of IVF treatments, the findings point to the existence of a `healthy patient effect' whereby the unhealthiest women in the population are deterred from pregnancy and infertility treatment.

Key words: complications of treatment/death/follow-up/IVF/maternal mortality


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
There are complications of IVF and related assisted reproductive technologies (henceforth referred to as IVF) which, though uncommon, are potentially life threatening. These include complications of ovarian stimulation, surgical procedures and pregnancy (Serour et al., 1999Go). Ovarian hyperstimulation syndrome (OHSS) is an iatrogenic condition that is most commonly seen in mild to moderate forms. In severe cases, it is characterized by massive ovarian enlargement, ascites, pleural effusion, haemoconcentration, oliguria, electrolyte imbalance and hypercoagulability (Schenker, 1999Go). The incidence of pulmonary and febrile morbidity associated with OHSS has been assessed in an Israeli multicentre study (Abramov et al., 1998Go, 1999Go). The death of an IVF patient due to OHSS has been reported from New Zealand (Cluroe and Synek, 1995Go).

There have been attempts to establish the frequency of complications of IVF treatment. Bergh and Lundkvist surveyed 12 IVF clinics in Nordic countries (Bergh and Lundkvist, 1992Go). Based on 10125 treatment cycles commenced, OHSS needing hospital care occurred in 0.7% of cycles. Complications of follicle puncture involving bleeding occurred in 0.5% of cycles with embryo transfer, while post-operative infections requiring antibiotics or surgical management occurred in 0.3% of cycles. Similar findings came from a retrospective analysis of 2495 IVF cycles at a single Dutch clinic (Roest et al., 1996Go). Hospital admission was required in 0.7% of cycles due to severe OHSS; adnexal torsion occurred in two patients, and both required ovariectomy; and post-operative infection requiring hospital admission occurred in 0.3% of cycles. Neither of these studies reported fatal complications of IVF treatment.

IVF pregnancies are at higher risk of serious complications because of the relatively high incidence of ectopic pregnancies, multiple pregnancies and advanced maternal age compared with normally conceived pregnancies in the general population. Unlike reports of IVF treatment complications, which have relied on voluntary reporting of adverse events in relatively small series of patients, data on IVF pregnancy complications have been collected systematically at a national level in Australia and elsewhere (de Mouzon and Lancaster, 1997Go). In the period 1979–97, there were 22 308 clinical IVF pregnancies in Australia and New Zealand, of which 4.0% were ectopic (Hurst et al., 1999Go), 28.2% occurred in women aged 35–39 and 6.2% in women aged >=40 years. In the same time period, 19.5% of IVF pregnancies of at least 20 weeks gestation were multiple pregnancies; of these 10.6% were higher order multiples. By comparison, general population data from 1997 show 12.9% of confinements in women aged 35–39, and 2.1% in women aged >=40 years. Multiple pregnancies made up 1.5% of all Australian confinements, of which 3.0% were higher order multiples (Day et al., 1999Go). Case reports of maternal deaths following IVF conception have appeared in the literature (Bewley and Wright, 1995Go; Sucov et al., 1995Go).

Although there have been estimates of the incidence of serious physical and mental complications of IVF treatment and pregnancy, there have been no systematic studies of the mortality rate in these patients. It is widely recognized that many patients experience psychological distress in the course of their infertility treatment, but it is not known whether they are at increased risk of suicide. Suicide rates provide a proxy measure for the most severe end of the spectrum of distress and have been used previously to examine associations with reproductive outcomes (Gissler et al., 1996Go).

We recently established a cohort of 29700 women who were referred for IVF treatment to at least one of 10 IVF clinics participating in our multicentre study. The cohort included women who received IVF treatment and others who were referred but not treated. The incidence of breast and gynaecological cancers in this cohort has been reported previously (Venn et al., 1999Go). Here, we present the results of record-linkage of patient names with the National Death Index (NDI), which allowed us to determine the mortality rate in this large and well-defined group of patients. Our aims were to compare the mortality rates of treated and untreated IVF patients with the mortality rate in the general female population, to determine the number of maternal deaths following IVF conception and to establish whether any deaths had occurred that could be related to complications of IVF procedures.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Study sample
The cohort consisted of 29 700 women who registered with at least one of 10 Australian IVF clinics before January 1, 1994. Details of the cohort study methodology have been described elsewhere (Venn et al., 1999Go). Clinics provided electronic data including patient's name, date of birth, address, date of registration with the clinic and dates and types of treatment. Data on IVF conceptions and pregnancy outcomes were not included in the electronic dataset. Women who commenced at least one treatment cycle, including natural cycles without ovarian stimulation, were classified as treated (n = 21086). Women classified as untreated were those who were registered for IVF but did not commence treatment (n = 8614). The reasons why women did not start IVF treatment were not routinely recorded, but included the occurrence of pregnancy while on the waiting list, pursuit of other treatment options, relationship or financial difficulties and change of mind for other personal reasons. The mortality rate of untreated women provided data for a comparison group of women with similar fertility problems and socio-demographic characteristics.

Ascertainment of deaths
Ascertainment of deaths from all causes was by record-linkage with the NDI for women residing in all Australian States and Territories except South Australia (n = 24 945). The NDI contains records on deaths that have occurred in Australia since 1980. Because of different privacy restrictions in South Australia, we were not able to link with the NDI for residents of this State (n = 4755). However, we were able to ascertain deaths from neoplasms in these patients by record-linkage with the State Cancer Registry. Record-linkage with the NDI and Cancer Registry was on name, date of birth, and State of residence. Cause of death was recorded by International Classification of Diseases codes, ninth revision (ICD-9) (World Health Organization, 1977Go).

When a death occurred within two years of an IVF treatment cycle, clinic records were checked to (i) verify the date of the last treatment cycle, (ii) find out whether the patient had conceived with treatment, and (iii) if the patient had conceived, to determine whether the death was classifiable as a maternal death. Maternal deaths were defined according to criteria used in previous reports on maternal deaths in Australia (NHMRC Maternal Mortality Working Party, 1998Go); that is, the death of a woman during pregnancy, childbirth or in the 42 days of the puerperium, irrespective of the duration and site of the pregnancy. This includes deaths from any cause related to or aggravated by the pregnancy or its management and deaths from incidental causes. This definition accords with World Health Organization (WHO) criteria used in the time period covered by this study, except in the inclusion of incidental deaths (World Health Organization, 1977Go). Deaths occurring >42 days but <12 months after the end of an IVF pregnancy were termed late maternal deaths.

Where a death occurred within 12 months of IVF treatment in a woman who had not conceived with treatment, and of causes that could be related to IVF treatment, further information was sought from the death certificate and from the IVF clinic records to investigate whether there appeared to be a causal relationship. Deaths occurring >12 months after IVF treatment were not investigated further.

Statistical methods
Standardized mortality ratios (SMR) were calculated from the ratio of observed to expected number of deaths. The time-age-specific rates for the Australian population for all-cause mortality, all neoplasms, breast cancer, diseases of the circulatory system, injury and poisoning and suicide were used with age in 5 year periods and time in four calendar periods (1975–1985–1986–1990–1991–1995–1996–1997) to calculate the expected numbers of deaths. Population rates were provided by the NDI.

The person-years of follow-up for untreated women were defined as those between registration in an IVF programme and death or end of follow-up (31.12.1997), whichever was first. The untreated group also included the person-years at risk accumulated by women who spent time on an IVF programme waiting list before commencing treatment. Person-years of follow-up in the treated group were calculated from the date of first cycle of treatment to death or end of follow-up.

The person-years were calculated using the DATAB module of the EPICURE software package (Version 1.8w). This apportioned the appropriate person-years into those treated with IVF and those not, and calculated the observed and expected numbers of deaths using the cohort data and the mortality rate data. Since data on deaths from causes other than neoplasms were not available from South Australia, numbers and person-years of follow-up are less for these causes. SMRs and 95% confidence intervals (CI) were then calculated using standard formulae (Clayton and Hills, 1993Go).


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The characteristics of patients in the treated and untreated groups, and the types of treatment cycles, are shown in Tables I and IIGoGo respectively for patients excluding residents of South Australia, since these patients contributed to analysis of deaths from neoplasms but not other causes. A total of 23.6% of the cohort registered with an IVF clinic before 1985. Women in the untreated group had a longer duration of follow-up than women in the treated group. In the early years of Australian IVF programmes, patients were less likely to commence treatment because of greater barriers such as long waiting lists, high costs and more burdensome treatment procedures.


View this table:
[in this window]
[in a new window]
 
Table I. Characteristics of patients* by IVF treatment status
 

View this table:
[in this window]
[in a new window]
 
Table II. Characteristics of treatment cycles*
 
In the treated group, 59681 cycles commenced of which 6528 (10.9%) were cancelled cycles where ovarian stimulation began but the cycle did not proceed to oocyte collection. The two most common drug regimens used for ovarian stimulation were clomiphene citrate and gonadotrophins, used predominantly until the late 1980s, or gonadotrophin-releasing hormone (GnRH) agonists and gonadotrophins, the preferred stimulation regimen in recent years. A total of 28.1% of women receiving ovarian stimulation had one or more cycles with clomiphene citrate and gonadotrophins, and 49.5% had one or more cycles with gonadotrophins and GnRH agonists.

Mortality compared with the general population
The observed and expected numbers of deaths in the IVF treated and untreated groups are given in Table IIIGo. Overall, mortality from all causes was significantly lower in IVF patients than in the general female population of the same age (72 deaths observed, 125 expected in the treated group; 51 deaths observed, 82 expected in the untreated group). The lower mortality rate in IVF patients, irrespective of treatment status, was apparent in the selected major cause of death groups and for suicide. There was little difference between the mortality rates in the treated and untreated groups for all-cause mortality, diseases of the circulatory system and injury and poisoning: the confidence intervals for the SMR estimates of the treated group overlapped the SMR point estimates in the untreated group. In the treated group, the number of breast cancer deaths was similar to the number expected from general population rates.


View this table:
[in this window]
[in a new window]
 
Table III. Observed and expected deaths by cause and by IVF treatment status
 
Deaths occurring within one year of the end of an IVF pregnancy
Maternal deaths are described in Table IVGo. Two deaths occurred within 42 days of the puerperium and two were late maternal deaths. The death rate, for deaths occurring within 42 days of the puerperium, was 3.4 (95% CI 0.9–13.8) per 100 000 IVF treatment cycles commenced or 11.7 (95% CI 2.9–46.8) deaths per 100 000 women treated.


View this table:
[in this window]
[in a new window]
 
Table IV. Deaths occurring within 12 months of an IVF pregnancy
 
Deaths occurring within one year of IVF treatment in women who did not conceive
Seven deaths occurred within 12 months of an IVF treatment cycle in women who did not conceive, as shown in Table VGo; two of these were due to diseases of the circulatory system, four from violent causes, and one due to neoplasm. For neither of the deaths due to circulatory disease was there any documentation of OHSS in the IVF patient records. The 36-year-old patient, who died of a subarachnoid haemorrhage, had had a frozen embryo thaw in an unstimulated cycle 80 days before her death: no embryos were transferred. Her most recent stimulated cycle was 6 months before her death. The death certificate noted the patient had had polyarteritis nodosa for 20 years causing renal failure. The 23-year-old woman, who died of a pulmonary embolism 32 days after transfer of two embryos, had five oocytes retrieved in that cycle. The death certificate noted disseminated intravascular coagulation and cardiogenic/septic shock. Although OHSS was not apparent, IVF treatment may have contributed to the development of the pulmonary embolism. The IVF clinic had had no further contact with this public patient and was unaware of her death. Based on one death in a non-pregnant patient, the treatment-related mortality rate was 1.7 (95% CI 0.2–12.2) per 100 000 treatment cycles, or 5.8 (95% CI 0.8–41.5) per 100 000 women treated.


View this table:
[in this window]
[in a new window]
 
Table V. Deaths occurring within 12 months of an IVF treatment cycle in women who did not conceive
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The overall mortality rate in this large cohort of IVF patients was significantly less than that in the general female population, taking into account age, duration of follow-up and calendar period. The lower mortality rate was seen in each cause of death group examined except breast cancer. The finding that deaths due to neoplasms, and breast cancer specifically, were no more common than expected is consistent with our previous findings in this cohort, showing no increase in the incidence of breast and gynaecological cancers (Venn et al., 1999Go). While this cohort represents a very large sample of IVF patients, too few deaths have occurred to permit accurate estimates of mortality rates for a broad range of specific causes.

The most likely explanation for this finding of a low mortality rate, which could reflect a `healthy patient effect', is in the selection factors that make IVF patients differ from the general population. The concept of a `healthy patient effect' comes from the healthy worker effect, initially observed in epidemiological studies of occupational diseases, which describes the phenomenon whereby workers exhibit lower death rates than the general population because of the tendency for the severely ill and disabled to be excluded from employment. We suggest that a healthy patient effect may arise in IVF programmes if the unhealthiest women in the community are deterred from planning a pregnancy or accessing IVF services. The suggestion that women who are planning a pregnancy may be healthier than other women is supported by findings of low mortality rates in pregnant and recently pregnant women compared with the general female population of the same age (Gissler et al., 1997Go; NHMRC Maternal Mortality Working Party, 1998Go; Ronsmans et al., 2001Go).

Studies of health differentials in the general population of women aged 25–64 years (Mathers, 1994Go) have shown that married women have lower mortality rates than single women; women in higher income families and those with higher education levels report better health; and socioeconomic advantage of area of residence is associated with lower mortality and better health. A relevant exception is the positive association between socioeconomic advantage of area of residence and mortality from breast cancer (Williams et al., 1991Go). These factors may have contributed to the low mortality rate in this IVF cohort, since the majority of women in this study were married (a legislative requirement for access to IVF in the State where most patients resided) and most IVF patients have private health insurance, which suggests higher socioeconomic status in Australia.

A further possible explanation for the finding of a low mortality rate in IVF patients is that we failed to identify some deaths, either because they were not recorded in the NDI, or because women appeared in the NDI under a different name. We would have had to miss 60 deaths, giving a record-linkage sensitivity of 67%, for the SMR for all-cause mortality to be not significantly <1.0. Linkage with the NDI was found to have 95% sensitivity in a recent large cohort study of Australian women (Powers et al., 2000Go). Our analyses of deaths occurring within 12 months of an IVF pregnancy or treatment cycle are unlikely to have been affected by ascertainment problems due to women changing their names.

The two maternal deaths that occurred within 42 days of delivery in our study sample had been identified in routine reports of national IVF statistics (Hurst et al., 1999Go). These reports relied on passive reporting of maternal deaths and it was reassuring, therefore, to see that record-linkage with the NDI did not identify any previously unreported deaths in this category. We were unable to calculate the maternal mortality rate as a proportion of all IVF confinements because we did not have complete data on pregnancies and confinements that had occurred in the whole cohort. However, from national data based on passive reporting up to 1997, there have been six maternal deaths in Australian IVF patients among 23 321 IVF and gamete intra-Fallopian (GIFT) confinements (Hurst et al., 1999Go), giving a maternal mortality rate of 25.7 per 100000 confinements. The overall maternal mortality rate in the Australian female population in the triennium 1991–93 was 10.9 per 100 000 confinements (NHMRC Maternal Mortality Working Party, 1998Go), which included direct, indirect and incidental deaths within 42 days of the puerperium. Advanced maternal age, a high proportion of multiple births and a high Caesarean section rate would be expected to contribute to an increased risk of maternal mortality rate in IVF pregnancies.

Among 17112 women in this study who had 59681 IVF treatment cycles, there appeared to be no deaths due to complications of OHSS. The finding provides some reassurance that appropriate clinical management of severe OHSS prevented patient deaths. IVF treatment may have contributed to the death caused by pulmonary embolism in a young woman 32 days after embryo transfer, though this could not be confirmed from available records. Venous thromboembolism and pulmonary embolus are the leading non-surgical causes of post-operative death (Laros, 1994Go). The reported incidence, 0.01–0.29% before and during pregnancy and 0.1–1.0% postpartum, varies widely and accurate diagnosis remains difficult. Pelvic surgery, even transvaginal oocyte retrieval, venous stasis and leg elevation are all known risk factors for deep venous thrombophlebitis and pulmonary embolism.

In conclusion, we found that IVF patients in this follow-up study had a low mortality rate compared with that in the general female population of the same age. The maternal mortality rate in IVF pregnancies appears to be higher than in the general population, probably due to increased maternal risk factors and the higher incidence of multiple births. Complications of OHSS were not directly related to any of the deaths observed. Treatment may have played a role in one death due to pulmonary embolism 32 days after embryo transfer. As well as providing some reassurance about the safety of IVF treatment, these findings indicate that selection processes are occurring whereby the unhealthiest women in the population are deterred from pregnancy and infertility treatment.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
This work was supported by grants from the Kathleen Cuningham Foundation, the Fertility Society of Australia, the Anti-Cancer Council of Victoria and IVF Friends. E.H. was supported by a La Trobe University Faculty of Health Sciences Overseas Visiting Fellowship. We are grateful for the assistance of staff at the Australian Institute of Health and Welfare and the South Australian Cancer Registry. We thank the staff at all participating IVF clinics, in particular J.Tyler (City West IVF), B.Bellinge (Concept Fertility Centre), S.Judd (Flinders Reproductive Medicine), J.Stanger (Lingard Fertility Centre), A.Speirs (Melbourne IVF/Reproductive Biology Unit), V.MacLachlan (Monash IVF), S.Junk (Pivet Medical Centre), R.Norman (Reproductive Medicine Unit, University of Adelaide), S.Sonneveld (Tas IVF) and H.Smith (Westmead Fertility Centre).


    Notes
 
6 To whom correspondence should be addressed. E-mail: alison.venn{at}utas.edu.au Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Abramov, Y., Elchalal, U. and Schenker, J. (1998) Febrile morbidity in severe and critical ovarian hyperstimulation syndrome: a multicentre study. Hum. Reprod., 13, 3128–3131.[Abstract]

Abramov, Y., Elchalal, U. and Schenker, J. (1999) Pulmonary manifestations of severe ovarian hyperstimulation syndrome: a multicenter study. Fertil. Steril., 71, 645–651.[ISI][Medline]

Bergh, T. and Lundkvist, O. (1992) Clinical complications during in-vitro fertilization treatment. Hum. Reprod., 7, 625–626.[Abstract]

Bewley, S. and Wright, J. (1995) Maternal death associated with ovum donation twin pregnancy. Hum. Reprod., 6, 898–899.[Abstract]

Clayton, D. and Hills, M. (1993) Statistical Models in Epidemiology. Oxford University Press, Oxford.

Cluroe, A. and Synek, B. (1995) A fatal case of ovarian hyperstimulation syndrome with cerebral infarction. Pathology, 27, 344–346.[ISI][Medline]

Day, P., Sullivan, E., Ford, J. et al. (1999) Australia's mothers and babies 1997. Sydney, Australian Institute of Health and Welfare, National Perinatal Statistics Unit.

de Mouzon, J. and Lancaster, P. (1997) World collaborative report on in vitro fertilization: preliminary data for 1995. J. Assist. Reprod. Genet., 14, 251S–265S.

Gissler, M., Hemminki, E. and Lönnqvist, J. (1996) Suicides after pregnancy in Finland, 1987–94: register linkage study. Brit. Med. J., 313, 1431–1434.[Abstract/Free Full Text]

Gissler, M., Kauppila, R., Merilainen, J. et al. (1997) Pregnancy associated deaths in Finland 1987–1994—definition problems and benefits of record linkage. Acta Obstet. Gynecol. Scand., 76, 651–657.[ISI][Medline]

Hurst, T., Shafir, E. and Lancaster, P. (1999) Assisted conception Australia and New Zealand 1997. Sydney, Australian Institute of Health and Welfare, National Perinatal Statistics Unit: 76.

Laros, R. (1994) Thromboembolic disease. In Creasy, R. and Resnik, R. (eds) Maternal–Fetal Medicine. W.B.Saunders, Philadelphia, pp. 792–803.

Mathers, C. (1994) Health differentials among adult Australians aged 25–64 years. Canberra, Australian Institute of Health and Welfare.

NHMRC Maternal Mortality Working Party (1998) Report on maternal deaths in Australia 1991–93. Canberra, National Health and Medical Research Council.

Powers, J., Ball, J., Adamson, L. et al (2000) Effectiveness of the National Death Index for establishing the vital status of older women in the Australian Longitudinal Study on Women's Health. Aust. NZ J. Public Health, 24, 526–528[ISI][Medline]

Roest, J., Mous, H., Zeilmaker, G. et al. (1996) The incidence of major clinical complications in a Dutch transport IVF programme. Hum. Reprod. Update, 2, 345–353.[Abstract/Free Full Text]

Ronsmans, C., Khlat, M., Kodio, B. et al. (2001) Evidence for a `healthy pregnant woman effect' in Niakhar, Senegal? Int. J. Epidemiol., 30, 467–473.[Abstract/Free Full Text]

Schenker, J. (1999) Clinical aspects of ovarian hyperstimulation syndrome. Eur. J. Obstet. Gynecol. Reprod. Biol., 85, 13–20.[ISI][Medline]

Serour, G., Rhodes, C., Sattar, M. et al. (1999) Complications of assisted reproductive techniques: a review. Assist. Reprod., 9, 214–232.

Sucov, A., Deveau, L., Feola, P. et al. (1995) Heterotopic pregnancy after in vitro fertilization. Am. J. Emerg. Med., 13, 641–643.[ISI][Medline]

Venn, A., Watson, L., Bruinsma, F. et al. (1999) Risk of cancer after use of fertility drugs with in-vitro fertilisation. Lancet, 354, 1586–1590.[ISI][Medline]

Williams, J., Clifford, C., Hopper, J. et al. (1991) Socioeconomic status and cancer mortality and incidence in Melbourne. Eur. J. Cancer, 27, 917–921.[ISI][Medline]

World Health Organization (1977) Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death. World Health Organization, Geneva.

Submitted on February 2, 2001; accepted on August 31, 2001.