The severity of the HIV/AIDS epidemic creates an urgent global demand for the expansion of available prevention, treatment, and care strategies, particularly those relevant to resource-poor settings.1,2 Clinical trials conducted at international research sites strive to maintain high ethical standards and comply with regulatory guidelines in an increasingly demanding regulatory environment.
Identification and reporting of adverse events (AEs) is an important aspect of compliance with the regulatory requirements and protection of human subjects. A focus on complete and timely reporting and appropriate management of AEs contributes to ethical conduct of HIV clinical trials. Compliance with AE reporting guidelines requires allocation of substantial human and financial resources, however. Prompt and accurate reporting of AEs for regulatory adherence is particularly challenging in resource-poor settings. When progression to AIDS and/or death may be more rapid amid conditions characterized by poverty, inadequate health services, poor nutrition, unsafe water, and endemic tropical diseases,3 more prevalent or severe HIV-related illnesses and complications can account for a considerably high AE rate. The cost of AE reporting is generally not explicitly estimated and is assumed to be built into the overall cost of trials, which may result in inappropriate resource allocation. Moreover, the widening scope of regulatory requirements, which are sometimes revised or reinterpreted during the course of ongoing trials, occasionally leaves research teams with an unfunded mandate.
We are not aware of any attempts to quantify the cost of AE-related activities in international clinical trials. An assessment of these costs should help in planning for resources, before the start of the trial and in response to postlaunch changes, and should facilitate the ethical conduct of clinical trials and compliance with regulatory requirements.
The aim of this study was to estimate the cost of identification, reporting, treatment, and follow-up of AEs in a perinatal HIV clinical trial in a developing country setting and to establish the relative cost for components of the AE reporting and management system.
The study was conducted from January 1, 2005 to June 30, 2006 at a large clinical research site in Kampala, Uganda. The Makerere University-Johns Hopkins University Research Collaboration (MU-JHU) has conducted HIV/AIDS research since 1988, and its active research program includes more than 15 previous and current studies addressing a broad spectrum of topics related to child and maternal health.
The AE system for an ongoing phase 3 clinical trial involving HIV-1-positive breast-feeding Ugandan women served as the framework for activity-based costing (ABC) analysis. This randomized controlled trial is not an investigational new drug (IND) study, and participants are randomized to receive HIV immune globulin (HIVIGLOB) or extended infant nevirapine (NVP) versus the standard NVP regimen alone for the prevention of mother-to-child transmission. Pregnant women were enrolled, and women and their infants were followed for 8 weeks and 18 months, respectively. The reportable period for AEs was 8 weeks postpartum for all study women and 14 weeks after birth for all study infants.
ABC is an established accounting technique that has been successfully applied to diverse settings in corporate and health care sectors.4-6 The relevant and reliable cost information it provides is used to enhance managed care and improve the efficiency of health care operations through the related practice of activity-based management (ABM). In contrast to traditional accounting systems, in which direct, indirect, and overhead costs are aggregated and assigned proportionally based on product volume produced or number of services provided,7 the strategy of ABC assigns costs more accurately by delineating specific services or procedures responsible for actual resource consumption.8
The definition of an “adverse event” included any health-related reaction, effect, toxicity, or abnormal laboratory result that had or could have had a deleterious effect on the well-being of a participant in the parent trial regardless of relation to study product(s). A “serious adverse event” (SAE) is defined as any experience that is fatal or life-threatening; permanently disabling; requires inpatient hospitalization; is a congenital anomaly, cancer, or overdose; or is otherwise judged to be serious by the on-site clinician. All illnesses, including background illnesses common in the research population, constituted AEs requiring study-specific documentation and reporting procedures. AEs occurring after the reportable period were still managed and treated by study staff. A diagnosing physician categorized AEs according to a severity grading system adapted from National Institute of Allergy and Infectious Diseases, Division of Acquired Immunodeficiency Syndrome (NIAID DAIDS) standard toxicity tables (April 1994).9 The relation of each event to study product was classified for NVP and HIVIGLOB.
The international trial represents collaboration with 6 organizations providing regulatory oversight. With 3 study arms, the protocol is complex and the AE reporting system is designed to be rigorous to ensure protection for all maternal and infant subjects involved.
The unit cost per AE was calculated by dividing the total costs of all AE-related activities by the number of identified AEs recorded in the clinical trial database. Costs for AEs attributed to study product or trial participation are indistinguishable from costs associated with standard HIV-related clinical care in this setting, because study procedures encompass protocol-related activities and provision of routine medical care for this population. Personnel cost per AE was further analyzed by severity and type of AE for a subset of directly observed clinic events.
From February 1, 2006 through June 30, 2006, time and motion studies were conducted prospectively to observe AEs identified in the MU-JHU clinic. Time and motion is a cost determination method requiring direct observation of staff and keeping track of time spent on various activities throughout the day.10 For study participants visiting the MU-JHU clinic, a comprehensive tracking sheet was generated to assess actual resource consumption based solely on performance of value-added components such as a doctor examination or blood draw. The actual duration for each task in minutes recorded by the responsible clinical staff member was multiplied by a salary rate average calculated for each personnel group and then summed to generate a total value reflecting personnel subcategory costs for identifying these AEs in the clinic.
Disease information pertaining to AE(s) diagnosed during the study visit was abstracted from the participant's medical file after being recorded by the examining physician. These time and motion studies traced activities limited to participant arrival at the MU-JHU clinic concluding with same-day departure. Subsequent procedures for AE tasks such as data form entry and verification, clinical tasks related to hospital admission, home visiting, and quarterly institutional review board (IRB) reporting, for example, could not be feasibly observed or directly followed with time and motion studies. There may be significant gaps in the process created by intervening days or weeks, and all AE-related duties are not performed in the clinic.
Therefore, to compute all costs for additional activities related to identifying, reporting, treating, and following up AEs, required resources were organized into component cost categories, including personnel, patient care costs, laboratory testing, equipment, and pharmacy expenses. Existing accounting and study data systems were reviewed, and the availability of cost and other clinical data was assessed. All accounting statements, with the exception of pharmacy reports, reflected actual expenditures in US dollars (USD). Cost inputs were quantified, and a complete list of accounting categories is detailed in the Appendix. Capital or overhead costs were not itemized in cost calculations, because expenses such as facility and equipment maintenance and utilities are generally incorporated into a study budget as part of the standard costs for conducting a clinical trial. Contributing categories were limited to costs for trial conduct incurred by the study itself and did not include expenditure of external resources from an individual or societal perspective.
The MU-JHU employed 330 staff, of whom 84 had duties directly related to AE activities. A standardized questionnaire was administered to these 84 staff to assess time spent on AE-related activities. Survey questions inquired about the number of hours or proportion of time an individual spent "on identifying AEs (eg, examinations, testing), on reporting AEs (eg, forms, reports), or on treating/following up AEs (eg, retests, reports). Study personnel deliver most primary health care to participants in an outpatient clinic setting, with additional support staff providing inpatient medical attention to mothers or infants requiring hospitalization.
For data entry personnel, the length of time elapsed while processing AE-related versus non-AE forms was tracked by database time stamps and provided an objective measure for determining AE-related personnel time. Similarly, the percentage of AE-related activities according to self-report by health visitors was verified against the proportional distribution from a weekly tally classifying health visits by purpose (eg, doctor requests; sick visits; maternal, neonatal, or infant deaths were designated as AE related).
Patient care includes expenditures such as meals for participants visiting the clinic, vehicle fuel for outreach efforts by health visitors, transportation compensation for participants visiting the research site, and financial assistance given to hospitalized participants. AE activity costs were allocated according to the proportion of unscheduled clinic visits versus all study visits for mothers and infants, conservatively assuming that unscheduled visits are attributable to illness and, presumably, AE related. Based on an economic life span of 6 years, vehicle costs were calculated in 2006 USD and AE-related expenses were derived by applying the percentage for AE-related personnel time self-reported by vehicle users. Study vehicles were used for escorting study participants home after enrollment to record place of residence, provision of transport in the event of illness or emergency, and home visiting by health visitors as a result of missed study visits or reported fatalities.
Contractual fees charged by the clinical laboratory for performing hematology, chemistry, microbiology, and immunology assays were used to measure all laboratory costs associated with the clinical trial. Consumable reagents, instruments, equipment, technical or administrative personnel, and other laboratory resources were not specifically itemized. Rapid laboratory testing (ie, statim [STAT] laboratory testing) was found during the pilot phase of participant observation to be universally correlated to AE identification, treatment, or follow-up; therefore, all STAT laboratory costs were collectively attributed to AEs. For all other requisitioned laboratory testing, the AE-attributable share of these ordered laboratory tests was based on a 3-month retrospective survey tabulating the frequency of each laboratory test for unscheduled versus scheduled study visits. Making the conservative assumption that scheduled appointments are routine study visits not associated with any AEs and any unscheduled visits are “sick” or AE related, an AE-related percentage for each type of test was assigned accordingly.
Costs for utilized equipment such as computers and printers reflected the distribution of AE-related personnel time reported by the primary user. The full cost of each workstation at the time of purchase was expensed, because equipment cost generally fell below the minimum threshold of $5000 USD for depreciation designated by institutional policy.
Pharmacy costs were provided by an internal accounting database, with monthly totals in Ugandan shillings. All prescribed medications are provided without charge to study participants, and pharmacy records reflect actual expenses billed to the study. Medication costs were uniformly grouped as AE associated, with the exception of specific vitamins or nutritional supplements provided as standard of care in the clinic.
For pharmacy expenditures that required conversion to USD, historical interbank currency exchange rates from OANDA Corporation (New York, NY) were averaged for each relevant month and applied.11
Statistical analyses were performed using Excel 2003 version 11.0.8033.0 (Microsoft Corporation, Redmond, WA). This analysis was considered to be exempt from review by the Committees on Human Research at the Johns Hopkins Bloomberg School of Public Health and approved by the Uganda National Council for Science and Technology and the HIV/AIDS Research Committee.
At the time of the study, a total of 688 women and 732 infants were being followed as active participants in the HIVIGLOB/NVP trial (HIVIG). Baseline evaluation of the HIV-positive pregnant women showed 458 cells/mm3 as the average CD4 cell count at enrollment and 424 cells/mm3 as the median CD4 count at enrollment. The median CD4 count for study infants at the 6-week study visit was 2213 cells/mm3, and it was 2134 cells/mm3 at 6 months. A total of 2709 maternal and 6319 infant AEs were recorded from January 1, 2005 to June 30, 2006.
The median of 9 AEs per study infant was calculated from birth beginning study enrollment until trial participation ended or censoring at the end of the study period. During the study period, 7.8% of the infants tested HIV-positive at birth. The median for mothers was 4 AEs per woman from study enrollment at 32 to 36 weeks of gestation until censoring or the end of the protocol's reportable period at 8 weeks postpartum.
Of the total 9028 events recorded in the clinical trial database, 4419 were considered mild (49%) and 3681 moderate (41%). Although 13% (839 of 6319) of infant events were categorized as SAEs, the overall occurrence of SAEs remained low at 11% (970 of 9028). With 1861 moderate AEs comprising the largest proportion (69%) of maternal complications, AEs for infants were less serious, with 59% reported to be mild (3714 of 6319). For infant AEs, 93% (5846 of 6319) were classified as not related or probably not related to the delivery of study product(s). Similarly, 96% (2606 of 2709) of maternal events were considered not or probably not related to study product(s) administration (Table 1A).
The most common maternal AEs were classified as respiratory (n = 459, 16.9%), obstetric (n = 418, 15.4%), and hematology/cardiovascular (n = 348, 12.9%). The most common AEs for infants were similar, with 1164 (18.4%) classified as respiratory, 932 (14.8%) as laboratory abnormalities, and 870 (13.8%) as hematology/cardiovascular (see Table 1B). The most frequent diagnosis for mothers was acute upper respiratory tract infection (URTI; 16.2%), followed by “other complications of pregnancy and childbirth” (8.2%) and neutropenia (6.9%). Infants were also most commonly diagnosed with acute URTI (n = 975, 15.4%), followed by increased creatinine (9.1%) and neutropenia (8.7%).
The unit cost per AE was calculated to be $101.97 after aggregating all expenses for clinical trial resources (Table 2). Of the $101.97, personnel was the highest contributor at $79.30 per AE (78%), followed by patient care expenditures of $11.96 per AE (12%). Laboratory testing accounted for $7.67 per AE (8%), with equipment and clinic medications contributing an additional $1.56 (2%) and $1.48 (1%), respectively. Overall, AE-related costs represented 32% ($920,581 of $2,834,692) of all study expenses assessed for the 18-month study period.
The global cost of $920,581 for AE-related use of study resources included staff member time, participant reimbursements, clinical laboratory services, information technology (IT) equipment, and medication expenses. Among the designated categories, personnel costs, including salary and benefits, accounted for 78% ($715,901) of total AE costs in this resource-limited setting. Resource utilization for AE activities as a percentage of total study costs varied widely for each category. Costs for medication dispensed to treat or alleviate maternal or infant AEs, for example, totaled a significant 79% of all pharmacy costs ($17,042) for the study period, whereas fees for AE-related laboratory testing comprised only 11% of the total $630,498 paid for all clinical laboratory analysis.
The proportions of AE-related personnel time by self-report also ranged considerably by job description (Table 3). Organized according to equivalent clinical, regulatory, or data management duties, study nurses dispensing medication reported spending 90% of their time engaged in AE-related tasks versus regulatory reviewers and medical officers, who reported 60% and 58% effort, respectively. Personnel costs for medical officers and various clinical staff accounted for 54% ($21,502 of $39 772) of all AE-associated personnel costs per month.
For time and motion studies, the identification stage of 516 AEs in the clinic was observed by randomly sampling 533 clinic visits (Table 4). Personnel costs during initial steps of the AE identification process were calculated to be $10.31 per AE. Personnel effort contributed by medical officers, clinic and dispensary nurses, counselors, health visitors, and regulatory reviewers for identifying 482 infant AEs was typically more costly ($10.59 per AE) than that required for the 34 maternal events ($6.30 per AE), which were detected in the clinic much more infrequently.
Most events (300 [58%] of 516) were detected during unscheduled clinic visits. Mild to moderate events (286 [55%] of 516) after the reportable period were predominantly observed. Overall, 75 different diagnoses were traced, with acute URTIs and diarrhea as the most commonly diagnosed AEs, representing 41% (209 of 516) of all conditions studied. Most (97%) of these occurrences (502 of 516) were deemed to be unrelated to delivery of either study product.
The use of ABC techniques with the implementation of prospective time and motion studies to assess study resource expenditure represents the first study to evaluate AE procedures quantitatively for a clinical trial in a developing country setting. Calculation of costs that underlie AE identification, reporting, treatment, and follow-up procedures at this international research site reveals that almost one third of study expenses are linked to the AE process. Although the cost of $101.97 per AE was calculated from a single HIV prevention trial, most costs were found to be overwhelmingly associated with most (93%) infant events and maternal events (96%) that were clinically unrelated to study product administration. Therefore, any variability in the cost for AE procedures may not be driven by the targeted research question or therapy regimen but influenced by the preexisting health status and/or underlying environmental conditions of the study population.
The relatively high proportion of AE-associated study expenses determined for this Ugandan perinatal trial is justified to conduct research involving human subjects responsibly in compliance with ethical and regulatory requirements. Researchers in developing countries typically provide routine clinical care to study participants, and documentation and monitoring procedures for AEs extend widely beyond the scope of a study's specific aims or delivery of an investigational product. Most trial participants rely solely on medical services and treatment provided at the research site by the study's clinical staff. Accurately understanding comprehensive costs and adequately providing resources should ultimately ensure improved planning and enhance safety for those enrolled in clinical trials.
Although personnel costs are generally low in developing countries, salary costs represent a major expense for conducting large-scale clinical trials. Large numbers of highly skilled individuals must be employed to implement rigorous procedures for a complex research protocol, and stringent guidelines must be followed for scientific reproducibility, regulatory adherence, and ethical propriety. Personnel expenses for a study are significantly greater than expenditure for clinicians in a local setting. Recruiting and training professionals such as pharmacists and midwives is costly, and additional instruction for good clinical practice (GCP), human subjects protection, or protocol-specific procedures is mandatory for research teams. Adequate compensation to retain qualified clinical data or regulatory staff remains worthwhile for researchers who strive to minimize loss to follow-up and value the consistency of clinical care provided to participants. Primary care delivered by highly specialized clinicians cannot be avoided when society's health resources are lacking and routine medical care cannot be separated from clinical research because study procedures integrate both seamlessly into 1 process.
Dedicating a sizeable portion of the study budget to employing staff is therefore not surprising, but the percentage of personnel time allocated to AE activities seems considerable. Calculating AE-related personnel effort based solely on those interviewed produced a conservative estimate. Those fulfilling support functions indirectly related to AE identification, reporting, treatment, or follow-up tasks were excluded from cost calculations, although the tasks they perform are vital. Clinical trial expenditures to bolster infrastructure by funding support staff may increase as protocols become more complex and comprehensive assessment or monitoring of patients remains a top priority.12
Similarly, time and motion studies provided a preliminary glimpse into some of the essential duties of clinic staff during the initial AE identification phase, but many tasks could not be captured or adequately traced during the observation period. Because incremental steps in the AE process may occur concurrently as multiple staff members perform related duties or with considerable overlap as reports are reviewed in batch, tracing an individual AE based on participant chart review is not feasible. Treatment and/or follow-up of AEs may also be an ongoing process with repeated lags spanning several clinic visits, and inclusion of this component restricted observation to newly diagnosed AEs to designate a standard starting point. With the average time for AE resolution reported to be 3 months in the clinical trial database, this prospective analysis documented only identification steps in the MU-JHU clinic, and the personnel cost of $10.31 does not encompass subsequent critical AE-related activities such as home visits or hospital admissions. The impact of these labor-intensive duties, which consume a significant portion of personnel resources, is underscored when comparing this with the total AE-attributable personnel cost of $79.30 estimated by record review.
For laboratory testing fees, approximation based solely on the frequency of tests ordered for unscheduled spontaneous visits may similarly underestimate AE costs. Any hematologic, chemistry, microbiology, or immunologic assays ordered during a scheduled study visit would be inadvertently excluded, although an AE may be suspected. Analysis of the time and motion AEs suggests that detection of an emergent AE may be equally probable for a study participant being evaluated during a scheduled routine visit versus an unscheduled or “sick” clinic appointment.
Furthermore, the minor percentage of laboratory testing costs calculated to be AE attributable may reflect an expanded emphasis on AEs. Screening assays may be incorporated into the protocol based on prior knowledge about potential side effects or AEs associated with the study product. Increased vigilance by investigators seeking to detect poor outcomes sooner or with greater sensitivity may result in a greater number of laboratory tests mandated as standard for study participants. According to this study design, the costs associated with these prescribed measures would be categorized as “routine,” and therefore not AE related. Clinical trial protocols generally mandate follow-up of any abnormal laboratory results irrespective of an AE's relatedness to study drug(s), and appropriate monitoring to document resolution requires rigorous oversight.
Among all AEs recorded in the clinical trial database during the study period and in the subset followed for time and motion studies, the most events were determined by the diagnosing clinician to be nonserious and unrelated to the delivery of study product(s). The observation that 2 of the top infant AEs (increase in creatinine and neutropenia) are laboratory abnormalities highlights the importance of standard reference values relevant for the local study population. Reporting guidelines relying on US-based tables to describe acceptable ranges may not be effective for classifying AEs or determining severity if standard values are markedly different in international settings.
The high incidence of AEs detected in this study sample of women and their infants is intrinsically linked not only to the prevailing maternal HIV infection but to elevated rates of disease morbidity found in developing countries. Evaluation, treatment, and resolution of these acute conditions, including URTI and diarrhea, nonetheless require substantial study resources, and the occurrence of these AEs remains admittedly greater in resource-poor settings, where formidable health challenges such as greater microbial burdens or widespread malnutrition have a notable impact.
Intensive identification and management of AEs helps to conduct clinical trials ethically in international settings but requires expenditure of substantial human and financial resources. Increasing expenses may be associated with provision of primary care for study participants, higher background rates of illness, and greater costs for highly trained personnel. Better understanding of these resource requirements should assist in improved planning and funding of international HIV-related clinical trials.
The authors gratefully acknowledge the principal investigators and HIVIG clinical staff at the MU-JHU Research Collaboration for their active support and invaluable participation in this project. They especially recognize the following individuals for contributing thoughtful input: Dr. Mary Glenn Fowler, Dr. Maxensia Owor, Joy Odongypiny, and Adrian Taylor.