1 International Journal of Medical Informatics (2005) 74, An electronic medical record system for ambulatory care of HIV-infected patients in Kenya Abraham M. Siika a,b,c, Joseph K. Rotich d, Chrispinus J. Simiyu c, Erica M. Kigotho c, Faye E. Smith b, John E. Sidle a,d, Kara Wools-Kaloustian a,d, Sylvester N. Kimaiyo a, Winston M. Nyandiko e, Terry J. Hannan f, William M. Tierney b,d, a Department of Medicine, Moi University Faculty of Health Sciences, Eldoret, Kenya b Regenstrief Institute, Incorporated, Indianpolis, IN 46202, USA c Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA d Department of Epidemiology, Moi University School of Public Health, Eldoret, Kenya e Department of Child Health and Pediatrics, Moi University Faculty of Health Sciences, Eldoret, Kenya f George Private Hospital, Kogarah, NSW, Australia Received 10 October 2004;accepted 3 March 2005 KEYWORDS Human Immunodeficiency Virus (HIV); Electronic medical records; International health; Acquired immune deficiency syndrome; Resource poor; Sub-Saharan Africa; Developing countries; Medical record system, computerized Summary Administering and monitoring therapy is crucial to the battle against HIV/AIDS in sub-saharan Africa. Electronic medical records (EMRs) can aid in documenting care, monitoring drug adherence and response to therapy, and providing data for quality improvement and research. Faculty at Moi University in Kenya and Indiana and University in the USA opened adult and pediatric HIV clinics in a national referral hospital, a district hospital, and six rural health centers in western Kenya using a newly developed EMR to support comprehensive outpatient HIV/AIDS care. Demographic, clinical, and HIV risk data, diagnostic test results, and treatment information are recorded on paper encounter forms and hand-entered into a central database that prints summary flowsheets and reminders for appropriate testing and treatment. There are separate modules for monitoring the Antenatal Clinic and Pharmacy. The EMR was designed with input from clinicians who understand the local community and constraints of providing care in resource poor settings. To date, the EMR contains more than 30,000 visit records for more than 4000 patients, almost half taking antiretroviral drugs. We describe the development and structure of this EMR and plans for future development that include wireless connections, tablet computers, and migration to a Web-based platform Elsevier Ireland Ltd. All rights reserved. * Corresponding author. Tel.: ;fax: / address: (W.M. Tierney) /$ see front matter 2005 Elsevier Ireland Ltd. All rights reserved. doi: /j.ijmedinf
2 346 A.M. Siika et al. 1. Introduction In 2001, the Departments of Medicine and Child Health and Paediatrics at Moi University, Eldoret and the Department of General Internal Medicine and Geriatrics at the Indiana University School of Medicine, in collaboration with the Moi Teaching and Referral Hospital in Eldoret, Kenya, established AMPATH: the Academic Model for Prevention And Treatment of HIV/AIDS [1,2]. AMPATH has the tripartite mission of patient care, medical education, and research focusing on HIV. It was the first organization to offer comprehensive ambulatory HIV/AIDS care in Kenya, including highly active antiretroviral therapy (HAART). Patient care also includes the management of opportunistic and other infections, nutritional and psychosocial support, prevention of maternal to child transmission of HIV using both HAART and Nevirapine, as well as diagnostic services. The AMPATH Training Institute (ATI) offers both didactic courses and mentored apprenticeships in the comprehensive care of the HIVinfected patient via multidisciplinary health care teams. AMPATH enrolled its first patient in November 2001 and has registered more than 4000 patients to date, almost half of who are currently on HAART. More than 90% of all registered patients have received therapy to prevent opportunistic infections. Prior to the conception of AMPATH, the Moi University-Indiana University partnership had implemented sub-saharan Africa s first ambulatory electronic medical record system in a rural Kenyan health center . Thought this system did not focus on patients with HIV/AIDS, it did demonstrate that the digital divide could be crossed, proving that an electronic medical record system could be maintained and could improve the efficiency and quality of care in a resource poor setting . This system now includes more than 150,000 records for more than 50,000 individual patients. Building on this experience and in keeping with the Institute of Medicine s recommendation to employ electronic medical records to improve the quality and lower the costs of patient care, the AMPATH medical record system (AMRS) was developed. The AMRS is sub-saharan Africa s first electronic medical record system for the comprehensive management of the clinical care of patients infected with HIV. This system, composed of both paper-based and electronic records, has led to uniformity in data collection and facilitated the retrieval of patient data for clinical care and research. 2. Methods 2.1. Sites AMPATH conducts special ambulatory HIV clinics in both urban and rural sites. The urban site is located at the Moi Teaching and Referral Hospital in western Kenya. Moi Hospital is Kenya s second national referral hospital and the site for its second medical school. It has a bed capacity of 400 and serves a population of more than 13 million people. As of July of 2004, AMPATH had also established HIV clinics at one district hospital (Webuye) and five rural health centers (Mosoriot, Turbo, Burnt Forest, Chulaimbo, and Amakura) all located within a 75 km radius of Moi Hospital supported with funding from the United States Presidential Emergency Plan for AIDS Relief (PEP- FAR). At Moi Hospital, AMPATH conducts a total of 10 half-day HIV clinic sessions per week in both the adult and pediatric HIV clinics. One third of all sessions are attended by both physicians and clinical officers (equivalent to physicians assistants in the USA). Clinical officers conduct the rest of the clinics but have access to an on call physician via a mobile telephone. The other sites conduct one to 10 half-day clinics per week, some attended by a physician and the others attended by clinical officers (physicians assistants) with fulltime access to physicians by mobile telephone. Two physicians one expert in adult medicine and one pediatrician, a clinical officer, and a nurse are on call to cover any emergencies during nights, weekends, and public holidays. All sites complete the same encounter data forms for each initial and return visit. Forms are collected once weekly and entered in the central database at Moi Hospital. In addition to receiving both primary care and HIV specialty care at all sites, all patients undergo nutritional assessments. Those found malnourished are either given nutritional support or nutritional advice. All patients on HAART or treatment for tuberculosis also receive drug adherence counseling at each clinic visit. At some sites, an outreach team conducts psychosocial support and counseling three times a weekly. There is also an active educational program in highyield organic farming (as most patients are subsistence farmers), that also provides nutritional support to the poorest patients, and a microenterprise development program that provides small loans to help families gain financial independence [1,2].
3 An electronic medical record system for ambulatory care of HIV-infected patients in Kenya Patients From a modest Pilot Program catering for 40 patients at the inception of the program, AMPATH has enrolled more than 4000 patients. To date, approximately 70% of who are cared for at Moi Hospital and 30% at the other seven sites. Approximately 85% of the patients are adults, and 60% are female. Individual philanthropy, foundations, the government of Kenya, and more recently the PEPFAR funds support patient care The structure and design of the AMPATH medical record system (AMRS) The AMRS consists of paper-based encounter forms and an electronic relational database (both software and encounter forms are available for free download at All encounter forms are completed by the nurse, physician, and/or clinical officer during each visit. Completed forms are transported physically to the AMPATH data center where they are entered by hand into the electronic central data repository. Internal audits found the data entry error rate to be less than 2%. After data are entered, the paper encounter forms are returned to the clinic and filed in patients paper charts The paper-based patient record The AMPATH HIV clinics (for both adults and pediatrics) use a standard eight page initial visit encounter form for new patients and a two page return visit encounter form for each subsequent HIV clinic visit. The majority of the data are entered via check boxes. This simplifies electronic data entry and also precludes problems encountered with illegible handwriting. The initial visit encounter form captures demographic and socio-economic data, marital status and family information, HIV exposure history, the HIV status of the patient s spouse, and whether the member of the couple has knowledge of the other s HIV status. A review of symptoms and signs is followed by medication history, including past and current use of HAART and other related medications. Information on alcohol use is collected via the alcohol use disorders identification test (3-item version) . Results of previous HIV-related tests are recorded, if known. A full review of systems and physical examination is carried out followed by staging of patients HIV infection using World Health Association (WHO) criteria . Eligibility for HAART and chemoprophylaxis for opportunistic infections are assessed and, if indicated, the relevant drugs are prescribed. There are sections for ordering tests (e.g., helper-t cells or tests of hepatic and hematologic complications of HAART), referrals to other clinics or hospitals, and recording the patient s next appointment. The return visit encounter form is shorter than the initial encounter form and records data necessary to chart the patient s HIV-related conditions and treatment The central data repository The unit of observation in the AMRS clinical data repository is the visit. Thus, the AMRS adopts a relational data structure consisting of various data tables that store different variables linked to each other by two fields: the patient s unique identifier and the visit date. In this case, the common field is the AMPATH patient number a unique number assigned to a patient upon enrolment at the HIV clinic. The structure of each table and the variables stored in each are shown in Table Summary flowsheet and computer reminders Prior research has shown that the earliest and perhaps most effective use of electronic medical record systems has been in the generation of structured data flow sheets which reduce the time to find relevant data and computer generated reminders [7,8]. Upon completing entry of data from each encounter form, the data entry technician generates a computer summary (Fig. 1) that contains relevant past history, vital signs, test results, and drug therapy in a collimated format that aids the user in discerning trends in the data. This is followed by patient-specific reminders (if indicated) for chemoprophylaxis against tuberculosis, Pneumocystis carinii pneumonia, and toxoplasmosis and follow-up testing of helper-t cell counts System design The AMRS was designed and implemented by programmers at Moi University and the Regenstrief Institute, Inc. at Indiana University. Prior experience in the development of the Mosoriot medical record system [3,4] guided the creation of this HIV-specific system. The Mosoriot system was focused on primary care and kept simple, with limited numbers of data fields describing comprehensive primary care delivered to all patients visiting a number of different clinics: adult medicine, pediatrics, antenatal, and family planning. The design and structure of the AMRS was influenced by its specialty focus of the clinical care on HIV/AIDS, the documentation requirements of funding programs such as MTCT-Plus , and minimizing the costs of
4 348 A.M. Siika et al. Fig. 1 Typical patient data summary report.
5 An electronic medical record system for ambulatory care of HIV-infected patients in Kenya 349 Table 1 Table name Registration Encounter HIV-related tests Other Tests HIV-related problem list Other Problems Past medical history Structure of the AMPATH HIV clinical database HIV exposure history Review of systems HIV drugs Description Stores the registration details of the patients such as the AMPATH Identifier, Name, Age, Clinic Site, Gender. Stores the patients clinical observations. Variables stored here include date of encounter, Temperature, Weight, blood pressure. Stores laboratory tests directly related to HIV such as CD4 cell count, CD4%, Viral Load, Absolute Lymphocyte Count, SGPT, SGOT. Stores data on the other tests that may not necessarily be HIV related such as urinalysis. It also stores data on HIV-related tests that may not have been pre-defined under the tblhivtests. Stores data on the HIV-related problems such as rash, oral thrush, etc. Stores data on problems that are not HIV-related such as malaria. Stores data on the patients medical history. This includes antiretroviral drugs that the patient has used in the past. Opportunistic infections and other relevant treatment done prior to the Clinic enrollment are also stored in this data table. Stores the patients clinic enrollment data particularly the pre-exposure history of the patient. Other variables stored include the marital status of the patient, partner s HIV status, sexual patterns of both the patient and their partners, number of children and their HIV status, Modes of feeding, etc. This is a review of the patients signs and symptoms such as fever, rash, night sweats, ear discharges, vision impairments, cough etc. Stores data relating to the patients treatment including Antiretroviral drugs, preventive and other therapy. designing, implementing, running, and maintaining the system. Important to both the Mosoriot system and the AMRS were (a) observing ongoing care and using local care providers to design a system that fit the care paradigm, rather than vice versa; (b) keeping the number of computers and other equipment at a minimum to enhance technical and financial sustainability;and (c) maintaining communication with the providers to evolve the system to meet their needs as the care system itself evolved. The HIV clinical data repository adopts a graphical user interface that closely follows the pattern of the paper encounter forms. Different subsets of data (e.g., exposure history, review of systems, physical examination) are navigated using tabs so that data can be entered into several hundred fields while keeping the entire data entry form on a single screen (Fig. 2). The programming is done using Microsoft s Visual Basic for Applications. The input fields are mostly pre-coded, meaning they can only accept valid field values as defined in the data dictionary. However, some fields, particularly those that require entry of new drugs that may not yet have a dictionary term, accept free text. Access to the system is restricted by the use of a password at the level of the computer and the MS- Access database. Similarly, access to the program code is restricted, making the system secure both from unauthorized access and unauthorized modification. There is a reporting function that generates reports on the number of patients enrolled by the various payment programs by site and patient age (adult or pediatric). Standard reports are generated for drug therapy and testing results. Custom reports are also available using queries and filters through MS-Access The data dictionary The heart of any electronic medical record system is the data dictionary (Fig. 3). It defines each data field and the pre-coded data that can be entered into each field. It contains a term number, term name, a reference term (for synonyms), term type (to define which data can be entered into each field), ICD-10 (International Classification of Diseases, Version 10) diagnosis codes, and (where applicable) the patient s charge for each test and treatment. Text describing each term is also included Special programs The AMRS has three special modules that store data for activities and initiatives that are critical to
6 350 A.M. Siika et al. Fig. 2 Data entry screen showing data collected from initial visit. AMPATH s clinical care mission. The first is the Maternal and Child Health Antenatal Counseling Module. This is a stand-alone sub-component of the HIV Care system. It is programmed in MS Access. The Table 2 The maternal and child health antenatal counseling module Field name Date ANC* No. PMTCT* Program No. Name Occupation Residence Counselor DatePreCounseled Result DatePostCounseled Remarks Description The date of encounter Patient assigned Antenatal Clinic number PMTCT registration number Name of patient Patient occupation Patient s physical address Name of counselor Date of pre-hiv* test counseling HIV test results Date of post-test counseling Any remarks by the counselor regarding the patient *ANC, Antenatal Clinic;PMTCT, prevention of mother-tochild transmission;hiv, human immunodeficiency virus. fields contained in this module are shown in Table 2. Two main outputs obtained from this system are: (a) The patient locator cards that are used for followup visits in case the pregnant patient tests HIV positive;and (b) Periodic reports showing the activities of the Antenatal Clinic. This report shows the crude HIV prevalence for antenatal mothers visiting the hospital. The second specialty AMRS module is the Food Inventory Module that manages the formula food used in the pediatrics alternative feeding program. It monitors the movement of stocks in and out of the AMPATH program s nutrition inventory. The input fields of this module are shown in Table 3. The main output from the Food Inventory Module is the electronic and printed version of a stock card showing the flow of stocks of formula food and the stock balances. The third specialty AMRS module is the HAART Pharmacy Module. This is an electronic spreadsheet acting as an inventory control system for managing the HAART pharmacy records. This module is a computer translation of the traditional bin cards that track drug movement between the
7 An electronic medical record system for ambulatory care of HIV-infected patients in Kenya 351 Fig. 3 Data dictionary. pharmacy, the clinics, and the patients. Data are used to manage pharmacy inventories and provide reports and requests to the agencies that fund HAART for AMPATH s patients. Key Input fields in the HAART Pharmacy Module are shown in Table 4. This system also produces a HAART adherence assessment report: helps in assessing the patient s adherence to drugs through drug refills, which are as accurate as and complimentary to the pill counts that are done at each visit [10,11]. Table 3 The maternal child health formula Food Inventory Module Field name Date QtyIn Name DateOfBirth Residence BreastFed QtyGiven Balance Description Date of transaction Quantity of stocks received Name of child Child s date of birth Physical address of the child Feeding history of the child: formula, breast-fed, mixed Number of tins given Number of tins in stock after issue 2.7. Program monitoring and evaluation As shown in Fig. 4, there has been a steady and sustained rise in the number of patients enrolling in the AMPATH HIV clinics. Data on patient enrolment are useful in planning for future needs for testing, drug therapy, supplies, personnel, clinic space, and other infrastructure needs. As previously mentioned, patients are assigned into different programs depending on their sociodemographic characteristics and vacancies within those programs (Table 4). Each program is unique with its own peculiar budgetary and staffing requirements. By electronically keeping track of patient recruitment and allocation, the AMRS plays an extremely important role in program harmonization. Every 2 weeks, the data manager makes a detailed report of the number of patients that have been registered. She also informs the clinicians about the programs that are filled up and those that still have vacancies, a necessary task where the need for care far outstrips the available resources. The AMPATH program makes use of standard treatment protocols. One way to determine the success of this program is by measuring the extent to which these set guidelines are followed. This is
8 352 A.M. Siika et al. Table 4 AMPATH HIV clinic patient enrollment by site and program (as of the end of July 2004) Funding program HIV clinic sites Total Moi Hospital Mosoriot Turbo Burnt Forest Webuye Chulaimbo Amakura Adults Awaiting assignment a MTCT-Plus b NASCOP c Philanthropy Self Pay Other Total adults Children Awaiting assignment a MTCT-Plus Philanthropy Self pay Other Total children All patients a Awaiting assignment = enrolled but not yet assigned to a funding program. b MTCT, mother-to-child transmission. c NASCOP, national AIDS STD control programme. done by querying the AMRS. For instance, 74% of adults in AMPATH care have had a baseline evaluation of their CD4 cell count of which 36% had a CD4 cell count <200 per ml. Only 72% of the patients qualifying for HAART had been started on ARVs and 53% of patients requiring preventive therapy for Pneumocystis carinii pneumonia were either on Cotrimoxazole or Dapsone. Less than a tenth (5.3%) of patients had been given Isoniazid preventive therapy. Similarly, certain laboratory tests form a baseline for each of the patients in the program while others are repeated at a pre-determined frequency. Using the AMRS, we are able to determine what proportion of patients care are adherent to these protocols. By periodically presenting this data to the HIV clinic clinicians, we are able to improve adherence. Program managers also make use of such data in planning for the allocation of scarce resources to the most urgent areas of the program Patient monitoring and evaluation Using the encounter forms enables the clinician to quickly but comprehensively go through HIV-infected patients commonest symptoms and Fig. 4 Initial AMPATH patient visits from November 2001 through July The solid line represents new patients seen per month. The broken line is cumulative patients enrolled in AMPATH.
9 An electronic medical record system for ambulatory care of HIV-infected patients in Kenya 353 Fig. 5 Reasons for missed clinic visits. Total number of missed visits in 27 months was clinical findings, aiding in identifying new onset opportunistic infections and other HIV-related conditions, reactions to drugs, and paradoxical reactions (immune reconstitution syndrome). Being a chronic illness without a cure, HIV care inevitably leads to many patient visits and diagnostic tests. The current AMPATH patient visit schedule dictates that all patients taking HAART be seen twice in the first month of initiating therapy and monthly thereafter. All other patients are given return visit appointments at the discretion of the clinician but no less frequent than 6 months. With time, patient visit paper-based records accumulate to the extent of making retrieval painstakingly slow and frustrating to the clinicians. The summary reports provide a means for quick access of such data and the ability to track trends over time thus facilitating early identification of toxicity, treatment failure and HIV complications. Because the summary report is given to the patient at the end of each visit, it becomes a portable patient medical record which can be provided to health care providers outside of the AMPATH system if the patient must seek care elsewhere. When resources allow it, future development of the AMRS will have a look-up feature whereby clinicians will search patients electronic records for specific data. AMPATH has a psychosocial support and outreach team comprising persons living with HIV/AIDS, many of who are on HAART. One of the major responsibilities assigned this team is promotion and enhancement of adherence to care (taking medication and clinic visits) among HIV-infected patients attending the AMPATH HIV clinics. The outreach team makes follow-up visits to the homes and workplaces of patients who have missed clinic appointments or have failed to pick up their HAART drugs. The outreach team is currently averaging 30 visits per month while the success of their visits stands at about 70%. The AMRS plays a key role in facilitating outreach work: It enables the outreach team to electronically track the patients clinic appointments and identify those who miss appointments and require a home visit by the outreach team. As patients on HAART are only supplied drugs that will last until the next clinic appointment, improving adherence to HIV clinic appointments is necessary in order to increase HAART adherence. For patients enrolled in the MTCT-Plus program , the database is able to create a linkage between the index woman and all other registered members of her household. This allows for the outreach team to inquire about the health and well being of other affected members of the family. The use of the AMRS for outreach activities has also led to a better understanding of why patients miss clinics. For example, in the first 27 months of operation, 2385 patients visited an AMPATH HIV clinic 13,443 times and did not keep 1761 appointments (12% of all scheduled appointments). As indicated in Fig. 5, the most frequent reason given for missing clinic visits was that patients were unable to get permission to leave work in order to attend
10 354 A.M. Siika et al. clinic. Using these data, AMPATH is considering the following options for the group of patients that are unable to attend clinic due to work: (1) open evening clinics (2) operate clinics on weekends and public holidays. For patients missing clinic because of lack of transport fee, AMPATH has established a welfare account for the most needy patients that will pay for their travel fee as well as supporting them financially (to buy food, pay school fees etc.) while they get back on their feet. These are just two tangible effects of AMRS data on patient care Preventing the mother-to-childtransmission (MTCT) of HIV Pre-test counseling for HIV testing is performed on all mothers attending the Moi Hospital Antenatal Clinic. Initially, only 30% of pregnant women agreed to HIV screening led to a change in testing policy. After adopting an opt-out strategy (testing unless the woman disagreed), antenatal HIV testing has approached 85 90%. By comparing these results with AMRS data, follow-up of HIV-positive woman was improved by guiding such patients to the HIV clinics for registration. The outreach group can then locate women who do follow-up in HIV clinic to reduce the number of HIV-infected women identified by the MTCT program fail to receive care in an HIV clinic. 3. Discussion We have taken our experience implementing a simple electronic medical record in a rural health center [3,4] and created a comprehensive electronic medical record system for the care of patients with HIV/AIDS. It balances the use of paper and electronic media;the need for simplicity in design and operation with the need to manage patients with a complex chronic illness;and the needs of individual clinicians caring for individual patients with the needs of those developing and managing a multisite, comprehensive HIV/AIDS treatment program. The AMRS uses a minimum of simple technology to aid in the maintenance and sustainability of the system in a resource-constrained environment. It is scalable and has expanded to manage data for eight clinics in four sites: one national referral hospital and three rural health centers. Yet, challenges remain. As the amount and complexity of the data have increased, clinicians, clinical managers, and researchers have begun to realize the potential of these data for managing and improving HIV/AIDS care. This puts enormous stress on the AMRS data management and analysis capabilities in Kenya, where there are insufficient numbers of trained personnel who are able to extract and analyze these data. Importantly, these hierarchical clinical data can be difficult to analyze for patient care and research. The unit of observation is the patient visit, but reporting and analysis is often needed at the level of the patient, provider, or site. Collapsing and summarizing these nested data is fraught with subtle biases , the most common of which is that such systems always know more about sicker patients (because they present more often for care). They are thus overly represented in the data and can make the population as a whole appear sicker than they really are. Managing data from an increasing number of geographically dispersed sites creates burgeoning problems in data management. Transporting completed encounter forms to and from remote locations where roads may be poor and at times impassible (e.g., during the rainy months) can cause data interruptions. Furthermore, increasing data needs for both clinical care and research can cause an exponential increase in database complexity that could cause errors in data storage and retrieval. As a result, the current platform of the AMRS (standalone computers) must eventually give way to a Web-based system with real-time uploading of data from remote sites into a central data repository and downloading of summary reports, etc. The current structure of the AMRS (a series of tables linked by patient identifier and visit date) must give way to a more sophisticated structure where the unit of record is the observation without regard to the type of observation: historical fact, vital sign, physical examination finding, drug prescribed, test result, etc. This is similar to the Regenstrief medical record system with which some of the authors have worked for more than 25 years . Unlike developing electronic medical record systems in developed countries, the lack of legacy electronic medical information avoids turf issues and the necessity of having to either evolve from existing to future systems or to link data from existing systems. On the other hand, in Kenya and other developing countries, there are no standards for data security and confidentiality. Developers of electronic medical record systems, such as the AMRS, in developing countries cannot wait for such standards to be established, nor can they assume that the standards in developed countries will be appropriate in developing countries. Development must continue nonetheless while keeping an eye on developing security and confidentiality standards and using common sense and the opinion of local
11 An electronic medical record system for ambulatory care of HIV-infected patients in Kenya 355 system users and patients to guide such development. 4. Conclusions The AMRS is the first functioning comprehensive electronic medical record system committed to managing and improving the quality and efficiency of care for patients with HIV/AIDS in sub-saharan Africa. It has played a significant role in patient care at all AMPATH sites. It has standardized patient data collection and made data retrieval much faster than the traditional paper-based record. It has enabled evidence-based decision-making for patient encounters and for the health system. The AMRS is affordable and represents a model system for recording critical HIV/AIDS data in resource poor settings that will be delivering an increasing amount of HIV care. It will allow those funding the rapid increase in HAART to know the return they are getting on their investment and hopefully encourage continued treatment of the worst medical disaster to ever befall humanity. Acknowledgments The authors would like to thank the following individuals and organizations for their role in creating, implementing, and supporting the AMRS: the AMPATH Co-Directors, Professors Haroun Mengech, MBChB, Director of Moi Teaching and Referral Hospital, and Barasa Otsyula, MBChB, Dean, Moi University Faculty of Health Sciences;the clinicians and staff in all AMPATH HIV clinics for their stalwart support of our efforts and their patients;and Simon Mungai for programming support and data management, Wyckliffe Chege and Elizabeth Ng ang a for providing ANC and outreach data. This work represents the opinions of the authors and does not necessarily represent those of the funding institutions supporting this work. Supported by grants from the Fogarty International Center, National Institutes of Health (number D43-TW0182), the MTCT-Plus Program, and the Bill and Melinda Gates Foundation. References  R. Volker, Conquering HIV and stigma in Kenya, JAMA 292 (2004)  J. Mamlin, S. Kimaiyo, W. Nyandiko, W. Tierney, Academic institutions linking access to treatment and prevention: case study, in: Perspectives and Practice in Antiretroviral Treatment, World Health Organization, Geneva,  T.J. Hannan, J.K. Rotich, W.W. Odero, D. Menya, F. Esamai, R.M. Einterz, J.E. Sidle, J. Sidle, W.M. Tierney, The Mosoriot medical record system: design and initial implementation of an outpatient electronic record system in rural Kenya, Int. J. Med. Inf. 60 (2000)  J.K. Rotich, T.J. Hannan, F.E. Smith, J. Bii, W.W. Odero, N. Vu, B.W. Mamlin, J.J. Mamlin, R.M. Einterz, W.M. Tierney, Installing and implementing a computer-based patient record system in sub-saharan Africa: the Mosoriot medical record system, J. Am. Med. Inform. Assoc. 10 (2003)  A. Gual, L. Segura, M. Contel, N. Heather, J Colom, AUDIT- 3 and AUDIT-4: effectiveness of two short forms of the alcohol use disorders identification test, Alcohol 37 (2002)  Scaling Up Antiretroviral Therapy in Resource-Limited Settings: Treatment Guidelines For a Public Health Approach, World Health Association, Geneva,  Q.E. Shiting-O Keefe, D.W. Simborg, W.V. Epstein, A. Warger, A computerized summary medical record system can provide more information than the standard medical record, JAMA 254 (1985)  S. Shea, W. DuMouchel, L. Bahamoud, A meta-analysis of 16 randomized controlled trials to evaluate computer-based clinical reminder systems for preventive care in the ambulatory setting, J. Am. Med. Inform. Assoc. 3 (1996)  A. Rosenfield, K. Yanda, AIDS treatment and maternal mortality in resource-poor countries, J. Am. Med. Womens Assoc. 57 (2002)  J.F. Steiner, T.D. Koepsell, S.D. Fihn, T.S. Inui, A general method of compliance assessment using centralized pharmacy records: description and validation, Med. Care 26 (1988)  P.W. Choo, C.S. Rand, T.S. Inui, M.L. Lee, E. Cain, M. Cordeiro-Breault, C. Canning, R. Platt, Validation of patient reports, automated pharmacy records, and pill counts with electronic monitoring of adherence to antihypertensive therapy, Med. Care 37 (1999)  W.M. Tierney, C.J. McDonald, Practice databases and their uses in clinical research, Stat. Med. 10 (1991)  C.J. McDonald, J.M. Overhage, W.M. Tierney, P.R. Dexter, D.K. Martin, J.G. Suico, A. Zafar, G. Schadow, L. Blevins, T. Glazener, J. Meeks-Johnson, L. Lemmon, J. Warvel, B. Porterfield, J. Warvel, P. Cassidy, D. Lindbergh, A. Belsito, M. Tucker, B. Williams, C. Wodniak, The Regenstrief medical record system: a quarter century experience, Int. J. Med. Inf. 54 (1999)