Methods

Using integrated MESH terms and keywords, we conducted a Medline search from January 1997 to December 2007. Exploded and focused MESH terms searches included combinations of Developing Countries, Safety, Quality Indicators (healthcare), Quality Control, Total Quality Management, Quality Assurance (healthcare), Quality of Healthcare, Data interpretation (statistical) and the keyword “measure$” (figure 1). The keyword search included intersections of results for safety or safety measures or safety or measurement or safety indicators or adverse events or medical audit or reporting system and developing countries or Africa or India or Brazil. This search strategy identified 530 articles; 29 articles that were not in English were eliminated.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Literature review methods.

The remaining 501 articles underwent title review. We eliminated 361articles where titles suggested an editorial or a focus on public health, road safety or occupational safety. Abstract review of the remaining 140 articles excluded articles that focused on one of the aforementioned topics or were not conducted in developing or emerging countries, defined here using the World Bank's list of low-income or middle-income countries.3 Furthermore, we excluded studies that examined the safety of a particular pharmaceutical compound.

After complete article review, we excluded five editorials, one statistical methods paper, and two general patient safety articles. We summarised the remaining 23 manuscripts and categorised them by study design and the type of measure examined in the study, using Donabedian's structure, process, outcome framework.4 For instance, structural measures might evaluate the number, training or qualifications of staff and the availability of required information or instruments; process measures might evaluate whether medications were safely administered, communications problems or adherence to guidelines; and outcome measures might include mortality or morbidity rates. Cohort studies were evaluated using the Newcastle–Ottawa Quality Assessment Scale for Cohort Studies.5 6 Results informed development of a plan for patient safety measurement in developing and emerging countries.

Results

We reviewed 12 prospective and 11 retrospective studies of patient safety. A summary of the literature, arranged by study design and summarising data collection methods and measurement type, is provided in the table 1. Two studies used structural measures, 13 used process measures and 16 used outcome measures to assess patient safety; several used more than one type of measurement.

Prospective studies

The prospective studies used three data collection methods to examine patient safety: continuous audit of patient or blood bank records, surveys, and direct observation of care processes and adverse patient outcomes. The three cohort studies were of modest quality, scoring six of nine stars on the Newcastle–Ottawa Scale.6

Structure, process and outcome measures were used to evaluate patient safety. The two studies that prospectively collected measurements of structure examined staffing, supplies and medical record documentation. At a single Indian hospital, researches investigated whether qualified practitioners were delivering injections in a safe manner.7 A South African study examined the quality of case notes by determining compliance with the Royal College of Surgeons of England case note guidelines.8 A Cambodian study of injection practices used process measures to investigate overuse of injections and improper disposal of injection waste.9

Six studies in Asian, African and South American countries used only outcome measures to assess patient safety. Two studies examined the frequency of adverse events related to transfusion10 and phlebotomy11; two examined adverse events in paediatric inpatients—one measured the rate and severity of adverse drug reactions in paediatric inpatient stays12 and the other evaluated the outcomes of different categories of adverse events that occurred during the transport of paediatric patients.13 A South African study measured the frequency of maternal near misses and mortality14; however, “near miss” was defined as any acute organ system dysfunction that could lead to death—this might be considered by most as an adverse event. The study did, however, use results to initiate quality improvement efforts. A Brazilian study examined postsurgical hepatitis C infection rates as an outcome measure.15

Three prospective studies used both process and outcome measures to assess patient safety. A Kenyan study examined performance measures and mortality rates to identify areas for improvement.16 A South African study measured administrative processes and common adverse outcomes leading to death to identify opportunities to reduce mortality in young children.17 Finally, a study in eight developing countries measured infection control processes and rates of nosocomial infection in intensive care unit settings to examine the impact of monitoring on infection and mortality rates.18

We found wide variation in the scope of prospective studies of patient safety. Some studies focused on specific patient safety issues such as transfusion error and unsafe injections, whereas others examined overall rates of adverse events and mortality at the hospital level. Only two studies were large-scale projects: one examined performance measures and mortality across 13 Kenyan government hospitals16 and the other audited 55 intensive care units across eight developing countries.18 Both demonstrate that coordinated, large-scale studies that examine patient safety outcomes are feasible in developing and emerging countries.

Retrospective studies

Most of the retrospective studies that we examined used record review to evaluate patient safety. Patient or case records, prescription records, admission and discharge records, as well as historical records of morbidity, mortality and reported adverse events were all used to estimate harm resulting from medical care. Surveys were rarely used. In several cases, direct observation of care processes served as a prospective complement to the retrospective outcome review. Both cohort studies scored six out of nine stars on the Newcastle–Ottawa Scale.6

Four retrospective studies used only process measures. One of these examined evidence-based intervention rates,19 whereas another estimated rates of adverse events using a national drug information database.20 A Brazilian study examined the process of infection control practices after a Salmonella outbreak,21 and a Chinese study measured unsafe injection rates and disposal of needles and syringes.22

Three of the retrospective studies that we identified used only outcome measures to evaluate patient safety. Studies in Zimbabwe23 and Nigeria24 examined neonatal morbidity and mortality; these papers identified preventive practices as opportunities to improve neonatal mortality outcomes. An Indian study asked physicians and pathologists to recall surgical pathology errors to estimate the actual rates of these errors.25

Four retrospective studies examined both process and outcome measures. Three of these studies evaluated the processes and resulting adverse events of specific medical interventions—anticoagulation,26 anaesthesia administration27 and transfusion practices.28 A Ghana study examined the relationship between the treatment process and mortality in patients with tetanus.29

The scope of retrospective studies was narrower than that of the prospective studies. Two studies were limited to the examination of a single disease at a single facility.21 29 Studies examining the safety of specific treatment processes22 26–28 or mortality rates for a particular age segment23 24 were conducted at single hospitals. Taking a somewhat broader perspective, a Peruvian study examined all patient safety-related interventions that were documented in the medical record over a 1-month period, albeit at a single healthcare facility; however, the study's measure of utilisation of evidence-based practices might be considered more in the realm of quality improvement than patient safety.19 None of the retrospective studies assessed patient safety across national borders.

Discussion

We conducted a review of the medical literature on patient safety in developing and emerging countries. We found a limited number of relevant articles that measured the structure, processes and outcomes of adverse events associated with medical care. About one-half of the studies were prospectively conducted; the other one-half were retrospective reviews of care. Most papers were limited to specific diseases or medical interventions at a single hospital. Studies focused on infection, adverse drug reactions, transfusion errors and injection errors; however, other common topics such as falls, wrong site procedures, suicide and pressure ulcers were not examined.

Despite the variation in the types and scope of measurement used for analysis, we found consistency in measurement methods: regardless of whether structure, processes or outcomes were the measures of interest, most studies used patient records as the source of data. Facility staff, individual researchers and trained reviewers performed these tasks. Clearly, the value of such chart audits will be dependent on the quality of the data in the charts. For countries with weak documentation systems, reporting of adverse events may be an appropriate initial step to begin to understand patient safety in emerging and developing countries. We believe this method would be preferable to the use of surveys or direct observation to collect data because both of those methods had substantial limitations: surveys relied on recall of adverse events, whereas direct observation was labour intensive.

There are several limitations to our review. First, our results are limited by the search methodology that we used. Lack of standardisation around patient safety terminology in the literature may have caused us to miss relevant articles. Second, our examination is likely tainted by publication bias in two ways: countries with the infrastructure to support and report measurement and those with the capacity to effectively write in English may be more likely than others to have been published. Such countries are likely to have fundamentals of patient safety such as clean water, waste disposal processes and professional certification processes in place. Finally, the measurement of patient safety relies heavily on patient medical records. To the degree that emerging and developing countries do not have fundamental structures necessary for evaluating patient safety—a unique medical record that includes regular annotation by staff and that reflects the care of a single, identifiable patient—those countries will likely have little success in measuring patient safety in any meaningful way. Despite these limitations, the measurements and methodologies found in the literature appear to be sufficient to inform the development of future metrics for patient safety in emerging and developing countries.

Recommendations

There has been substantial progress towards developing measures of patient safety that can be monitored over time in developed countries. However, many of these measures are predicated on features of developed countries' healthcare systems that are not likely to exist in emerging and developing countries, such an established patient record, administrative databases that can be exploited for research purposes, organisation that lends itself to evaluation and oversight, and a research infrastructure. Although some might suggest that aggregate patient safety outcomes in emerging and developing countries might be revealed using national measures of infant or maternal mortality rates, such measures may not create a tension for change at the sharp end of healthcare delivery, where improvement in patient safety occurs. Furthermore, lack of availability of patient records may make such aggregate data suspect.

Given the limited resources available in developing countries, the reliance of most established measurement activities on patient charts and the weakness of many documentation systems, we believe that a focus on structural and process-oriented aspects of patient safety may be the most efficient way to improve patient safety in emerging and developing countries. For instance, the adoption of several structural and process-oriented fundamental safe practices—such as establishing safe medical waste disposal, using safe needle practices and using unique patient identifiers that are attached to patient charts—may be effective first steps in ensuring the basics of patient safety and might facilitate the measurement, monitoring and improvement of patient safety through proxy indicators. Important steps to improve patient safety in emerging and developing countries would be adoption of a model of patient safety and implementation of fundamental safe patient practices (such as a unique identifier for each patient, standardised documentation of medical treatment, implementation and documentation of safe medication administration processes, and credentialing of healthcare providers). These fundamental building blocks can build a robust documentation system that can serve as the basis for patient safety measurement in emerging and developing countries.

Our review of the measurement of patient safety in developing and emerging countries was encouraging in one respect: it is evident that patient safety evaluations can be conducted in these countries. Moreover, we found that when charts are available, chart audit is a promising method for monitoring patient safety. Establishing fundamental safe patient practices, integrating those processes into routine health services delivery and developing patients' expectations that such processes be present are necessary prerequisites to progress towards safe patient care in emerging and developing countries.