Cost of Fertility Treatment in Women of Different Ages
Cost of Fertility Treatment in Women of Different Ages
A total of 2463 women from the Aberdeen City District attended the fertility clinic at the AFC for the first time between 1998 and 2008. Of 1736 (70.5%) women whose BMI data were available, 56% had a normal BMI (i.e. 18.5–24.9 kg/m), 25% were overweight (i.e. 25–29.9 kg/m) and 16% were obese (≥30 kg/m). Table I shows the baseline characteristics of women in different BMI categories along with the treatments received. Women with a BMI <18.5 kg/m were significantly younger than women with BMI in the normal range and women with BMI between 30 and 34.9 kg/m. Over half of all women with BMI ≥35 kg/m presented with anovulatory infertility whilst more women (36%) in the normal BMI group had unexplained infertility. Fewer women with BMI over 35 kg/m underwent IVF/ICSI treatment as compared with women with lower BMIs. Women with missing BMI data were significantly older (33 versus 32 years of age), had more secondary infertility (55 versus 41%), smoked more (27 versus 22%) and had a higher percentage of tubal factor infertility (26 versus 17%) compared with women with complete data (Supplementary data, Table SVII http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). A significantly lower proportion had anovulatory (14 versus 23%) and unexplained (24 versus 31%) infertility. They also had less clomifene citrate (14 versus 21%), gonadotrophin (1 versus 3%) and IVF (29 versus 34%) treatment.
Of all the women referred to AFC from primary care with a diagnosis of infertility during the study period, 1258 (51.1%) had a live birth within 5 years with 694 (55.1%) of these being treatment-independent (spontaneous) conceptions. In those ≤30 years of age, 36.7% conceived spontaneously, compared with only 9.1% of women over the age of 40 years. Out of the 1211 (49.2%) women who received any treatment, 564 (46.6%) had a treatment dependent live birth while 164 (13.5%) had a subsequent spontaneous live birth. Out of the 1252 (50.8%) women who did not receive fertility treatment, 530 (42.3%) had a spontaneous live birth (Table II). In women of all age groups, spontaneous conception rates were higher than those as a direct result of active treatment (Table II).
Table III highlights the mean investigation, fertility treatment and pregnancy costs by age and BMI group. Generally, the average cost of investigations tended to decrease with age and also tended to be lower in women classified as severely obese (≥35 kg/m). The average treatment costs generally appeared highest in women with normal BMI, except in the youngest age group. Average pregnancy and neonatal costs followed a similar pattern (reflecting the higher live birth rate in women with normal BMI) but in some age groups these costs were highest among women with BMI >35 kg/m. The total costs of investigation and treatment were highest among women who were 30 years or younger, with BMI <25 kg/m.
Table IV shows the number of women in different age and BMI groups who were investigated, along with the resulting costs. In comparison with younger women with normal BMI, fewer women who were older (>40 years) or heavier (BMI >30 kg/m) underwent a laparoscopy, possibly due to concern about increased surgical and anaesthetic risks. Cheaper investigations, such as blood tests (e.g. mid-luteal progesterone and other hormonal tests), were more frequent in the obese group up to the age of 40 years. With regards to fertility treatment (Table V), fewer women in the older (and also the youngest) age groups, and in the higher BMI groups, received IVF treatment. By contrast, more women in the youngest age group (and within some age groupings the higher BMI groups) received treatment with clomifene citrate.
To further explore the relationship between costs of fertility investigation and treatment with age and BMI, we estimated these costs whilst adjusting for factors associated with the treatment pathway and treatment success. The modelling information (including the parameter estimates for the cost model) is contained in Supplementary data, Table SVIII http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1. The parameter estimate for a particular factor affecting outcome is defined as the predicted increase in cost associated with a one unit increase in the value of that covariate.
The predicted costs from these adjusted analyses are presented in Table VI for a cohort of women with unexplained infertility—a group without an absolute barrier to conception who would be expected to have a reasonable chance of treatment-independent pregnancy. The results show a cost increase associated with treatment which is higher among women in the lowest BMI group (across all age groups), and also highest among women aged 31–35 years, followed by women aged 36–40 years (compared with women in the youngest and oldest age groups).
A similar approach also assessed the predicted probability of live birth. The live birth outcome model shows an uplift in the predicted probabilities of live birth with exposure to treatment, which is fairly consistent across age and BMI categories (~10%), except in the oldest age group where a slightly smaller increase in the probability of live birth is observed (see Table VI and Supplementary data, Table SIX http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). Table VII shows the difference in costs and the difference in the probability of live birth between treated and untreated couples with unexplained infertility across different BMI and age groups. The ratio of these two quantities represents the additional cost per additional live birth associated with fertility treatment. This ratio appears to be fairly consistent across the three youngest age groups. However, it is consistently higher in women over the age of forty than it is for women in the other age groups (across all BMI categories). For example, in women over 40 years of age with a BMI between 18.5 and 25 kg/m, the cost of an additional live birth with treatment was £32,785.52. For a woman aged 36–40 years with similar BMI, the cost of an additional live birth with treatment was £24,249.10. A surprising finding is that this ratio tends to fall as BMI increases within all age groups. Similar findings were observed for the baseline risk predicted costs and live birth probabilities (see Supplementary data, Table SX http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1).
In both models the year of registration was statistically significant meaning that the costs and live birth outcomes changed over time, as one may expect. To investigate this further we split the cohort into two time periods, i.e. 1998–2003 and 2004–2008, and refitted the models for each. Generally, there was little difference between the effects of age, BMI and treatment in the two models (supplementary data, Tables SXI and SXII http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). For the cost model, the interaction between age and treatment status was not statistically significant for the earlier time period but was for the latter time period, with significantly less costs for treatment in women over 40 years.
Multiple imputation of missing data (mainly BMI) did not appear to alter the magnitude and direction of the results substantially (Supplementary data, Tables SXIII and SXIV http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1) but the inclusion of 947 extra patients increased the statistical power resulting in narrower confidence intervals and more highly significant parameter estimates. The predicted probabilities of live birth were generally slightly higher across all the age and BMI categories than those based on the analysis on the complete data (Supplementary data, Table SXV http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). The predicted costs from the analysis with imputed data were also slightly higher for all age and BMI categories apart from the lowest BMI category which had lower estimates.
Results
Patient and Treatment Characteristics
A total of 2463 women from the Aberdeen City District attended the fertility clinic at the AFC for the first time between 1998 and 2008. Of 1736 (70.5%) women whose BMI data were available, 56% had a normal BMI (i.e. 18.5–24.9 kg/m), 25% were overweight (i.e. 25–29.9 kg/m) and 16% were obese (≥30 kg/m). Table I shows the baseline characteristics of women in different BMI categories along with the treatments received. Women with a BMI <18.5 kg/m were significantly younger than women with BMI in the normal range and women with BMI between 30 and 34.9 kg/m. Over half of all women with BMI ≥35 kg/m presented with anovulatory infertility whilst more women (36%) in the normal BMI group had unexplained infertility. Fewer women with BMI over 35 kg/m underwent IVF/ICSI treatment as compared with women with lower BMIs. Women with missing BMI data were significantly older (33 versus 32 years of age), had more secondary infertility (55 versus 41%), smoked more (27 versus 22%) and had a higher percentage of tubal factor infertility (26 versus 17%) compared with women with complete data (Supplementary data, Table SVII http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). A significantly lower proportion had anovulatory (14 versus 23%) and unexplained (24 versus 31%) infertility. They also had less clomifene citrate (14 versus 21%), gonadotrophin (1 versus 3%) and IVF (29 versus 34%) treatment.
Live Birth Outcome
Of all the women referred to AFC from primary care with a diagnosis of infertility during the study period, 1258 (51.1%) had a live birth within 5 years with 694 (55.1%) of these being treatment-independent (spontaneous) conceptions. In those ≤30 years of age, 36.7% conceived spontaneously, compared with only 9.1% of women over the age of 40 years. Out of the 1211 (49.2%) women who received any treatment, 564 (46.6%) had a treatment dependent live birth while 164 (13.5%) had a subsequent spontaneous live birth. Out of the 1252 (50.8%) women who did not receive fertility treatment, 530 (42.3%) had a spontaneous live birth (Table II). In women of all age groups, spontaneous conception rates were higher than those as a direct result of active treatment (Table II).
Costs of Investigation, Treatment and Pregnancy
Table III highlights the mean investigation, fertility treatment and pregnancy costs by age and BMI group. Generally, the average cost of investigations tended to decrease with age and also tended to be lower in women classified as severely obese (≥35 kg/m). The average treatment costs generally appeared highest in women with normal BMI, except in the youngest age group. Average pregnancy and neonatal costs followed a similar pattern (reflecting the higher live birth rate in women with normal BMI) but in some age groups these costs were highest among women with BMI >35 kg/m. The total costs of investigation and treatment were highest among women who were 30 years or younger, with BMI <25 kg/m.
Table IV shows the number of women in different age and BMI groups who were investigated, along with the resulting costs. In comparison with younger women with normal BMI, fewer women who were older (>40 years) or heavier (BMI >30 kg/m) underwent a laparoscopy, possibly due to concern about increased surgical and anaesthetic risks. Cheaper investigations, such as blood tests (e.g. mid-luteal progesterone and other hormonal tests), were more frequent in the obese group up to the age of 40 years. With regards to fertility treatment (Table V), fewer women in the older (and also the youngest) age groups, and in the higher BMI groups, received IVF treatment. By contrast, more women in the youngest age group (and within some age groupings the higher BMI groups) received treatment with clomifene citrate.
Relationship Between Costs and Age and BMI Groups
To further explore the relationship between costs of fertility investigation and treatment with age and BMI, we estimated these costs whilst adjusting for factors associated with the treatment pathway and treatment success. The modelling information (including the parameter estimates for the cost model) is contained in Supplementary data, Table SVIII http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1. The parameter estimate for a particular factor affecting outcome is defined as the predicted increase in cost associated with a one unit increase in the value of that covariate.
The predicted costs from these adjusted analyses are presented in Table VI for a cohort of women with unexplained infertility—a group without an absolute barrier to conception who would be expected to have a reasonable chance of treatment-independent pregnancy. The results show a cost increase associated with treatment which is higher among women in the lowest BMI group (across all age groups), and also highest among women aged 31–35 years, followed by women aged 36–40 years (compared with women in the youngest and oldest age groups).
Additional Cost Per Additional Live Birth Associated With Treatment
A similar approach also assessed the predicted probability of live birth. The live birth outcome model shows an uplift in the predicted probabilities of live birth with exposure to treatment, which is fairly consistent across age and BMI categories (~10%), except in the oldest age group where a slightly smaller increase in the probability of live birth is observed (see Table VI and Supplementary data, Table SIX http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). Table VII shows the difference in costs and the difference in the probability of live birth between treated and untreated couples with unexplained infertility across different BMI and age groups. The ratio of these two quantities represents the additional cost per additional live birth associated with fertility treatment. This ratio appears to be fairly consistent across the three youngest age groups. However, it is consistently higher in women over the age of forty than it is for women in the other age groups (across all BMI categories). For example, in women over 40 years of age with a BMI between 18.5 and 25 kg/m, the cost of an additional live birth with treatment was £32,785.52. For a woman aged 36–40 years with similar BMI, the cost of an additional live birth with treatment was £24,249.10. A surprising finding is that this ratio tends to fall as BMI increases within all age groups. Similar findings were observed for the baseline risk predicted costs and live birth probabilities (see Supplementary data, Table SX http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1).
In both models the year of registration was statistically significant meaning that the costs and live birth outcomes changed over time, as one may expect. To investigate this further we split the cohort into two time periods, i.e. 1998–2003 and 2004–2008, and refitted the models for each. Generally, there was little difference between the effects of age, BMI and treatment in the two models (supplementary data, Tables SXI and SXII http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). For the cost model, the interaction between age and treatment status was not statistically significant for the earlier time period but was for the latter time period, with significantly less costs for treatment in women over 40 years.
Multiple imputation of missing data (mainly BMI) did not appear to alter the magnitude and direction of the results substantially (Supplementary data, Tables SXIII and SXIV http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1) but the inclusion of 947 extra patients increased the statistical power resulting in narrower confidence intervals and more highly significant parameter estimates. The predicted probabilities of live birth were generally slightly higher across all the age and BMI categories than those based on the analysis on the complete data (Supplementary data, Table SXV http://humrep.oxfordjournals.org/content/29/10/2199/suppl/DC1). The predicted costs from the analysis with imputed data were also slightly higher for all age and BMI categories apart from the lowest BMI category which had lower estimates.
Source...