Economic Impact of Abortion Related Morbidity and Mortality

Economic Impact of Abortion Related Morbidity and Mortality:

Modeling Worldwide Estimates

Michael Vlassoff

30 April 2006

1. Introduction

Abortion-related morbidity and mortality (ARMM) impacts welfare at the individual, household, community and national levels. Out of an estimated 46 million induced abortions that take place every year in the world, around 19 million are unsafe abortions.[1] More than 6 million of these result in serious medical complications that require hospital-based treatment. Of these cases, many suffer long-term effects, including an estimated 1.5 million women who annually suffer secondary infertility. The costs that these figures imply is a matter of importance for public policy. This paper takes a step toward understanding these costs by developing a framework for quantifying the economic impact of unsafe abortion.

The complications from unsafe abortion have been listed elsewhere, for instance by Bernstein and Rosenfield (1998) and WHO (1995). Empirical studies on abortion complications, however, show that the list of complications is very long. Annex Table 1 attempts to organize this long list into immediate complications, later complications and “other complications,” which are reported only sporadically in the literature. What is evident from this table is that a complete costing of abortion-related complications would need evidence on the prevalence of all complications. Since much of these data are missing, the costing exercise undertaken here is perforce limited to the more prevalent complications.

The paper is divided into four substantive sections. In the next section, direct costs are examined. These are the costs which must be met for the adequate treatment to ARMM cases, whether paid for by the state, by households or jointly. In the second and third sections, indirect costs are investigated. These are economic costs borne by individuals, households and countries resulting from death or disability consequent to unsafe abortions, mediated through lost income or lower productivity. The fourth section summarizes the different estimated costs in the light of the limitations of data and the assumptions that were made, pointing out priorities for future research into ARMM costing.

2. Direct Costs to the Health System

PAC Interventions

Post-abortion complications cover a very wide range of medical problems. The treatments and interventions mentioned in the empirical literature, however, are less extensive. Annex Table 2 provides a list of treatments referred to inpublished empirical studies of PAC.[2] The following is a summary of the medical procedures and treatments reported on in this literature:

Operative Procedures

Colpopuncture

Colpotomy[3]

Dilation and curettage

Hysterectomy

Intestinal resection

Laparotomy[4]

Manual vacuum aspiration

Resuscitation, intensive care unit

Surgery (unspecified)

Other Procedures

Blood transfusions

General anesthesia

Intravenous antibiotics

Intravenous fluids

Local anesthesia

Sedation

Medicine Administered

Abortifacients

Analgesics

Antibiotics

Antimalarial drugs

Flagyl

Hematinics[5]

Tetanus vaccination

Vitamins

This list is incomplete. For instance, treatment for poisoning, renal failure, psychosis, infertility, inter alia, would require interventions not listed here.

Although it would be desirable, in calculating the cost of PAC, to start out from a set of inputs such as those listed above, there are as yet so much missing detailed information that this type of “bottom up” approach is problematic.[6] Nonetheless, in Annex III anexploratory exercise using the WHO Mother-Baby Package is undertaken which reflects the “bottom up” methodology, as a contrast to the cost estimations which follow. This approach uses a “top down” methodology in which costs per case – which encompass the totality of inputs – form the starting point for cost calculations.

Data Considerations

Several empirical studies have provided cost estimates of PAC. An extensive literature search found 27 such costing studies in which an overall cost-per-case estimate of PAC was provided. These studies are summarized in Table 2.A.[7] Seventeen countries and seven (out of 15) United Nations sub-regions for the developing world are represented in the table and just two countries, Mexico andNigeria, account for seven of the 31 studies. In Table 2.A, average costs are presented in two ways. First, all average costs were converted into US dollars at the time of data collection.[8] Then, costs were calculated in US dollars for 2005, adjusting for inflation from the time of the study to 2005.[9]

It can be observed that costs vary greatly from one study to the next, the range being from $4.33 to $504 (in 2005 US dollars). Even studies in two broadly comparable countries using a similar methodology may yield very different results: Fonseca et al. (1997) estimated cost per case at $19 in Brazilwhile Billings and Benson (2005) reported a cost per case of $282 in Mexico. Even within the same country large variations are seen in the cost estimates: average cost ranges from $59 to $282 in Mexico and from $96 to $424 in Nigeria. It seems obvious that the definitions of what constitutes costs must be varying widely from one setting to another.[10] A number of possible explanations could be advanced to explain these differences. Personnel time costs may be estimated according to the actual patient-provider contact time in one study but by dividing the salary cost of personnel by the number of patients attended in another study. Indirect costs, such as overhead costs, capital depreciation costs, administrative costs, etc., may by included in one study but excluded in another.

Probably the most significant source of variation in costs, however, is the inclusion or exclusion of subsidized costs from study to study. It is scarcely conceivable that PAC costs in Mexico, for instance, could be 15 times the costs inBrazil, two countries with similar social and economic settings. Quite likely the Mexican cost estimate includes many more of the real costs for post-abortion treatment than does the Mexican estimate because the latter cost is highly subsidized, hiding many of the true costs from easy detection. To arrive at reasonable regional or global cost per case estimates, therefore, it will be necessary to make some assumptions about the inclusion/exclusion of indirect costs, which are equally valid components of the true cost of PAC even if they may be more difficult to measure.

To get a better appreciation of cost measurement issues, we look briefly at two important components of PAC that are often reported in costing studies, namely, hospitalization and blood transfusions. Table 2.B presents all costing studies that had specific findings regarding hospitalization of patients as part of PAC. The simple average length of stay (ALOS) across all studies is 86.2 hours, or a little over three days. Again, there is a very wide variation in hospitalization, from 9.9 hours in a Ecuadorian study to over 26 days in a Nigerian study.[11]Sample designs differ across these studies. Although all studies purported to observe women coming to hospitals for treatment of complications after experiencing induced abortions, there may be some studies that covered only the most severely complicated cases or report from settings where the overall safety of abortion procedures is extremely low, while other studies may cover settings where abortion methods are on average less dangerous.

This possibility is reinforced when we divide the available studies into those reporting on operations research (into the introduction of MVA to replace D&C as the preferred technique for evacuating the uterus of products of conception) and those that are not. Of the 37 studies reporting hospitalization data, 25 studies were of the former type and 12 were not. The average stay for the MVA-D&C comparison studies was 35 hours,[12] whereas in the remaining studies the average was 153 hours.[13]The former studies typically select women who first trimester abortions in order to hold gestation constant in comparing the two procedures; other studies represent a cross-section of women with post-abortion complications.

The relationship between average length of stay and average cost is direct, as we would expect. That is, as length of hospitalization increases, so does average cost.[14] It is clear, therefore, that in estimating the global cost of ARMM due consideration should be taken of escalating cost per case as complications become more severe.

Table 2.C summarizes findings from 14 PAC costing studies that contained data on blood transfusions, an important component of PAC costs. In these studies, on average about 8 percent of women seeking care after induced abortions received blood transfusions. The amount of blood given to women who had transfusions is more difficult to estimate since it often was not reported. The two most recent studies estimated that 1.3 liters were administered per case, whereas the older studies (reported in Fortney, 1981) estimated that around 0.6 liters was given per woman. It should be noted that some of these studies mentioned that blood transfusions were restricted by the availability of blood in hospitals. The lower figure may therefore be more indicative of supply constraints than of effective demand.

The data presented in Tables 2.A, 2.B and 2.C clearly show that a wide range of PAC costsare reported in the literature and that cost variability persists within regions and even within single countries. We make the assumption here that studies reporting low costs did so because important indirect costs were not measured. To estimate a global average cost per case of PAC, we therefore take the average of the ten highest estimates from Table 2.A, which likely have included cost components omitted in other studies. This average is $177 (US dollars, 2005).[15]Another approach – categorizing studies by type of sampled cases – was considered, but in any case the average costs per treatment from that approach were quite close to the approach adopted.[16]

In order to arrive at disaggregated estimates of costs per case by region and severity, a useful categorization of the severity of post-abortion complications developed by Rees et al., 1997, was employed. The three categories are:

Severity
Category / Symptoms
Low / Temp. ≤ 37.2°C
No clinical signs of infection
No system or organ failure
No suspicious findings on evacuation / and
and
and
Moderate / Temp. 37.3 – 37.9°C
Offensive products
Localized peritonitis / or
or
Severe / Temp. ≥ 38°C
Organ failure
Peritonitis
Pulse ≥ 120
Death
Foreign body/mechanical injury on evacuation / or
or
or
or
or
Source: Rees et al., 1997, p. 433

Data on abortion-related complications by severity are limited. One study in South Africa (Kay et al., 1997) has used this categorization to estimate PAC costs according to severity of complication. Table 2.D presents some of that study’s main results, showing average costs (converted into US 2005 dollars) by severity category for three hospital settings in South Africa. As can be seen, costs increase by severity and from district to tertiary hospitals. As this is the only available study that provides estimates at this level of detail, its results are used in the present paper to estimate global PAC costs by severity level (see Annex II below). These estimates in turn are then disaggregated by developing region. The disaggregated costs per casecalculated in this way are shown in Table 2.E. This aspect of the estimation methodology is explained in detail in Annex II.

Methodology

In the estimations that follow an attempt is made to summarize direct costs to health systems for treating the complications from unsafe abortions. These are divided into the following direct costs:

• Direct costs resulting from women hospitalized for post-abortion complications
• Direct costs that would result if women, who need hospital-based treatment but receive no treatment, were in fact to obtain treatment
• Direct costs to women with less severe complicationsneeding/receiving care at the primary health care level
• Direct costs that would result if women suffering infertility due to unsafe abortion received adequate treatment.

As mentioned above, no attempt is made to apportion costs according to who pays, the individual, the household or the state. However responsibilities are assigned, these costs are borne by society at large. It should also be noted that some of these costs are real and some are notional. For instance, an estimated four million women with serious complications after unsafe abortions do not receive the hospital-based care they need. The costs of providing adequate medical services to these disenfranchised women should be taken into account in estimating total ARMM costs because such services should be provided even if they are not at present.

Health System Cost Estimates

Out of approximately19 million unsafe abortions per year, about 9.6 million result in complications serious enough to need hospital-based treatment.[17]Of these, an estimated 5.6 million are actually admitted to hospital while the remaining 4.1 million are not treated for a variety of reasons. Another approximately onemillion women have less severe complications that can be treated at the primary-health-care level. Table 2.F displays these numbers disaggregated by the 15 regions of the developing world.[18]The proportions by severity level are maintained constant across all regions for lack of more specific regional data. This limitation probably distorts regional prevalence estimates. For example, only 19 percent of all serious complications in the developing world are estimated to occur in sub-Saharan Africa, even though around 43 percent of all maternal deaths due to unsafe abortion come from that region. When more detailed regional data become available this deficiency will be able to be addressed.

Singh, 2006 has estimated rates of hospitalization for abortion complications in developing regions. Using these rates, it is estimated that 5.6 million women in developing countries were hospitalized in 2005 for post-abortion complications.[19]However, many other women in need do not get treated: “Based on surveys of key informants in all three major regions of the world, it is estimated that an additional 15-25% of all women having abortions experience medical complications but do not obtain care for these complications” (Singh, 2006). Using this range of percentages, we estimate that the number of untreated serious post-abortion cases in 2005 amounted to around 4.1 million women in the developing world.[20]

Combining the disaggregated costs per case from Table 2.E with the numbers of cases given in Table 2.F produces estimates of total costs to health care systems at the secondary and tertiary levels. These costs are shown in Table 2.G, upper panel. The total cost for treatment in a hospital setting for the developing world as a whole is estimated at almost one billiondollars for the 5.6 million women who actually obtained hospital-based treatment. Of this total, 70 percent would be spent on treatments for complications of moderate severity which include, inter alia, evacuation requiring admission as an inpatient and use of a regular operating theatre (Kay et al., 1997). As seen in Table 2.G, more than one quarter of the total would be expended in South-central Asia. Sub-Saharan Africa’s share of the total amounts to about 26 percent.

The lower panel of Table 2.F shows hypothetical costs if the 4.1 million women, who need hospital-based care for post-abortion complications but presently do not receive it, were to obtain such care.[21] In total, another $707 million would need to be spent on their treatment. In many contexts, infrastructure to treat these extra case is currently inadequate, so added capital costs for construction and training would also be required.

Other Direct Costs

Besides the costs to health systems for treatment of the estimated 5.6 million women receiving care in a hospital setting, there are another million women or so who suffer from minor complications that can be treated at the primary health care level or privately. Pain management, treatment for anemia and counseling are typical treatments that could be delivered at this level of care.

Unfortunately, no empirical study was found that had cost data in this area. In lieu of better data, we use a widely recognized reproductive health costing model developed by the WHO, the Mother-Baby Package (MBP). This software package has standard default values (which were calibrated based on conditions in rural areas of Uganda) that can be used to obtain a proxy value of the cost per case for post-abortion women needing primary health care. The MBP models basic reproductive health care at three service levels including the primary level (“health posts”). In its standard formulation the package estimates a cost per client of US$12.55 for an array of maternal and newborn services delivered by health posts such as antenatal care, family planning services, etc.[22] This figure is used as a proxy for cost per case for PAC for women with minor complications.

The estimate costs by region are shown in Table 2.H. The total cost for treating minor post-abortion complications in developing countries is estimated at $12.5 million for 2005. These costs may be borne by the public health care system if primary health care is provided without charge to all women while in other settings the costs may represent out-of-pocket expenses to women or households in settings where such care is provided privately. In other situations, these costs might be shared between a partially subsidized public health system and private contributions.