Barriers to Accessing Emergency Medical Services in Accra, Ghana: Development of a Survey Instrument and Initial Application in Ghana

Most respondents thought the number of ambulances insufficient and said they would rather use a taxi—perceived to be faster—in a medical emergency. Nevertheless, people generally had favorable attitudes of existing public ambulance services, although few knew of the toll-free emergency number and many thought it appropriate to use ambulances to transport corpses. Targeted public education, along with improved capacity of ambulance agencies to handle increased caseload, could improve use.

BACKGROUND E mergency medical services (EMS) are a community's gateway to acute and emergency medical care for members of the public facing timesensitive, condition-critical illness and injury. 1,2 When implemented appropriately, EMS systems are an effective, frontline, public health intervention to reduce the disproportionately high morbidity and mortality in low-and middle-income countries. 1,3,4 The formation of locally appropriate EMS systems in low-resource settings, to provide emergency care and transport, has been promoted by international bodies, such as the World Health Organization and the African Federation for Emergency Medicine. 1,2,5,6 Thus, EMS systems, regardless of their state of development, play a critical role in the continuum of ensuing medical care. 1,2,7 Emergency care usually begins in the community, when someone identifies a perceived emergency condition and attempts activation of the local EMS system. This ideally triggers a cascade of events resulting in a timely response of expertise, resources, and service directed to patient stabilization and/or safe emergency patient transportation to the nearest appropriate facility. 1,2,5,8,9 The current norm in many low-and middle-income countries is, however ironic, to use a private vehicle or a taxi to transport the injured or ill person to the hospital, even when EMS has an active presence in these communities. 10,11 Across Africa, in-hospital and prehospital emergency care systems are being developed to serve diverse, multicultural, and multilingual populations of varying socioeconomic strata. 2,6,12,13 For example, innovative programs in which motorbikes equipped with stretchers are activated in Malawi to transport emergency obstetric patients, and Ghana's National Ambulance Service (NAS) provides professional crews and time-sensitive emergency transportation for patients. 9,14,15 Ghana boasts one of sub-Saharan Africa's thriving EMS systems. Formed in 2004, the NAS is comprised of a fully operational ambulance fleet with 160 basic life support-equipped ambulances and more than 1,200 emergency medical technicians (EMTs). It has a nationwide operational footprint, providing free emergency services to the citizens of Ghana. Within each of Ghana's 10 regional capitals in 2013, NAS had at least 1 ambulance station and several ambulances staffed by a crew of trained EMTs. Accra, the national capital, had the largest NAS complement of ambulances and personnel, with 8 ambulances and approximately 100 EMTs. 16,17 Still, the Accra ambulance-to-population ratio is approximately 1:250,000-a ratio that is 5-10 times below expert-recommended ratios for lower-income countries-and the mean response time is about 18 minutes. 16,17 Injuries are the most common reason for public utilization of NAS services. 16,17 Despite sustained growth of NAS over the past decade, annual reports have indicated low public utilization, which may be a contributor to continued poor outcomes of acute care. The reasons behind low utilization have not been studied sufficiently. 9,16,17 This study was undertaken to characterize and quantify the range of barriers-from the demographic and psychosocial to the financial, knowledge, and cultural barriers-that prevent Ghanaian citizens in Accra from appropriately accessing local EMS resources. In Accra, we hypothesized low public opinion of ambulance services, poor knowledge of ambulance access, and thus a strong public preference to use non-ambulance means of transport during medical emergencies.

Conceptual Approach
Pechansky and Thomas provide a conceptual framework of 5 dimensions that affect patient access to or entry into health care systems: availability, accessibility, accommodation, affordability, and acceptability (Box). 18 Applications of this framework have elucidated causes for increased mortality pertaining to accessing emergency care, for emergency conditions such as sepsis and respiratory and obstetric emergencies. 19 The domain of content and relevant elements for our survey instrument were identified by a multidisciplinary research team, including experts in emergency care, EMS, African public health, and survey design, following discussions and a review of the literature. Desired attributes of the survey instrument included comprehensiveness, simplicity in execution, standardized oral administration, and ease of comprehension by respondents, independent of their sociopolitical, cultural, and geographic background. The survey was designed to be delivered on a tablet computer (Nexus 7, Google Inc., USA) using survey software (droidSurvey 2013, Google Inc., USA) to facilitate data collection. The research team developed survey questions for each element. Most of these questions were structured to enable quantitative categorization of responses. Due to the exploratory nature of this research, openended questions were also included to promote less-restricted responses and provide a richer explanation of issues raised by respondents.

Development and Validation of Survey Instrument
The research team defined the domain of the survey as all citizen-perceived barriers potentially impeding EMS access in low-resource sub-Saharan African settings. Relevant elements to the domain were screened in the context of Pechansky and Thomas's domains of access (i.e., availability, accessibility, accommodation, affordability, and acceptability).
The survey instrument was written in English and designed to be orally administered to respondents by trained bilingual research assistants in English (the official language), French, or Twi (the most dominant local language). Each research assistant received more than 20 hours of training on the sampling technique and survey instrument, which included completing 15 supervised surveys each. The survey contained 8 main sections: Of the 114 total questions, respondents were asked a maximum of 104, while 10 questions pertained to survey administration. The instrument had skip logic successfully embedded within it.
Early versions of the instrument underwent multiple rounds of revisions and sequential stages of validation. Three external survey experts provided feedback on its structure, including the syntax, grammar, and meaning of the chosen questions. Cognitive testing of a preliminary version of the instrument was performed in the United States (in Atlanta, Georgia, and in Ann Arbor, Michigan) and then in Ghana (in Accra and Kumasi) among 19 community members in total. During cognitive testing, feedback was provided on the meaning and clarity of questions, brevity, and ease of understanding and of responding to the survey. Pilot testing on the tablet computer was also conducted in Ghana, between April and June 2013, among 30 community members to ensure trouble-free data collection, to assess survey duration, and to resolve any impediments to conducting the survey.
According to the 49 community members and the 6 expert reviewers who participated in both phases of pilot testing and qualitative critique of the survey, the survey instrument passed face validity and content validity, and it was deemed

BOX. Five Dimensions of Access to Health Care
Availability: the relationship of the volume and type of existing services (and resources) to the clients' volume and needs.
Accessibility: the relationship between the location of supply of services (or resources) and the location of clients.
Accommodation: the manner in which the services (or resources) are organized to meet the needs of clients and clients' perceptions of the appropriateness of the way services are organized.
Affordability: the relationship between the cost and perceived value of services and the clients' ability to pay.
Acceptability: the relationship of the clients' perceptions and attitudes toward the service (or resources) to the actual characteristics of the service, as well as to the perceptions and attitudes of providers toward certain clients.

Source: Penchansky and Thomas. 18
Global Health: Science and Practice 2015 | Volume 3 | Number 4 sound with regards to construct validity and translation. To test the instrument's consistency when administered by different surveyors, the instrument was tested by multiple surveyors over 2 days: 2 survey administrators on day 1, and 3 survey administrators on day 2. Interrater reliability measurements of key variables yielded Krippendorff's alpha scores between 0.66 to 1, suggesting good to excellent interrater reliability. 22,23 Intrarater and test-retest reliability were non-applicable because the EMS educational intervention (an integral part of the survey mission) would skew respondents' subsequent answers and introduce error into tests of reliability. All initially stated desired attributes were satisfied by the final design of the instrument.

Sampling
Surveys were conducted in a representative crosssection of the general public in Ghana's capital city, Accra, from June to August 2013. Accra is a coastal city in southern Ghana, and the metropolitan area had a population of approximately 1.8 million in 2012, distributed over 173 square kilometers. 24,25 The rationale for selecting Accra was that the Accra metropolitan area has the highest concentration of NAS ambulances and the highest population volume and density in Ghana. 26,27 Therefore, Accra qualified as a high-yield survey study site in which to reach an initial understanding of barriers to EMS access in the urban Ghanaian population.
To conduct balanced community-based random sampling, we replicated the population survey technique used in the 2003 and 2008 versions of the Ghana Demographic and Health Survey. 28 In this approach, the Accra metropolitan area was divided into 11 geographic clusters, termed enumeration areas (Figure). To ensure valid statistical analysis could be conducted within each enumeration area, a target sample size of approximately 40 in each of the 11 areas was chosen, giving an overall sampling size of 440.
Two research assistants (JB, CA) were trained by the study's lead investigators. The research assistants were fluent in English (both JB and CA), French (JB), and Twi (CA). In order to attain as broad a sample as possible, diverse areas including the road side, commercial areas, residential areas, schools, places of worship, and recreational areas were used as recruitment sites. By sampling every fifth person within each recruitment site and by using a goal of 6 to 8 interviews per site, an approximately equal number of people were interviewed within each enumeration area.

Inclusion and Exclusion Criteria
All persons 18 years of age and older, located within the 11 enumeration areas during our survey period, were eligible to participate after providing informed consent. Respondents who did not speak English, French, or Twi were excluded.

Data and Statistical Analysis
After we conducted the survey, we exported the data from the droidSurvey database into Microsoft Excel (2007, Redmond, WA). Quantitative data from closed-ended questions were cleaned and coded. Qualitative data from open-ended questions were reviewed by the authors and coded individually, into themes. Following initial coding, 2 authors (NM and SR) compared codes, discussed and resolved discrepancies, and categorized the codes into thematic areas. Missing data were noted as such, and no imputation was performed for missing variables. Cleaned and coded data were exported into a statistical software program, Stata (Stata Corp, College Station, TX), for analysis. Descriptive statistics were performed for quantitative data, using frequencies, means, and standard deviations (SDs), as appropriate.
In the survey, participants were presented with 2 hypothetical emergency situations and asked how they would prefer transport of the victim to the hospital. In the first scenario, the participant witnessed a pedestrian on the street struck by a car, implying accessibility to many private and commercial vehicles. In the second scenario, a relative was burned badly in a house fire, implying limited access to private or commercial vehicles. The primary outcome for regression analysis was the response to the question, ''If you saw a pedestrian hit by a car and they needed to go to a hospital immediately, how would you get them to the hospital?'' Answers were categorized, then dichotomized to ''ambulance'' (combining the NAS ambulance and other ambulance options) and ''non-ambulance,'' then were analyzed using logistic regression.
An analytic model was developed to assess the effect of independent variables on this primary outcome variable. Independent variables in our model included respondent's age, sex, education level, prior experience with ambulances, knowledge of local ambulance services, knowledge of appropriate ambulance functions, and knowledge of the local emergency access number, all of which were assessed as part of the survey. To explore the relationship between independent variables and the outcome variable (decision to call an ambulance), descriptive cross tabs with chi-square analysis were performed. Variables were selected according to previous literature and the expert opinion of the authors. Variables that were found to be significant in bivariate analysis at the 0.1 level were entered into a multivariate model.

Ethical Approval
Ethical approval was obtained from the following human subjects committees: Colorado Multiple Institutional Review Board, University of Michigan Institutional Review Board, and the Ghana Health Service Ethical Review Committee.

RESULTS
Of 597 people who were approached, 468 completed the survey, yielding a survey response rate of 78.4%. Almost all (490%) persons who refused participation did not give a reason for why they declined to consent. Respondents were sampled from all 11 enumeration areas, with between 38 and 57 respondents in each area. The majority of the respondents were surveyed in business locations (n = 204, 43.6%), followed by the street (n = 135, 28.8%) and residential FIGURE. Survey Locations Within the 11 Demographic and Health Survey Enumeration Areas for Accra, Ghana locations (n = 89, 19.0%). The remainder (n = 33, 7.1%) were surveyed in locations coded as ''other,'' mainly postgraduate schools, such as law school. Location data were missing for 7 participants (0.85%). Most of the surveys were conducted in English (n = 273, 58.3%), but a substantial proportion were translated (n = 195, 40.2%), virtually all in Twi (n = 184 in Twi, n = 1 in French, n = 10 not documented).

Background Demographic Characteristics
Slightly more than half of the participants who completed the survey were men (n = 248, 53.0%) ( Table 1). The mean age was 34.9 years (SD, 12.9), with the majority (n = 277, 59.2%) of respondents between 18-35 years. The large majority (n = 448, 95.7%) resided in the Accra metropolitan area. Only 53 participants (11.3%) owned a car, but 453 (96.8%) owned a cell phone. About one-third (n = 159, 34.0%) of the respondents had attained post-secondary education.

Prior Ambulance Use
In the preceding 5 years, 350 (74.8%) respondents had had a personal experience and/or had witnessed an experience with either or both a medical (n = 246, 57.7%) or traumatic (n = 205, 47.1%) incident that required acute or emergency care. Of those prior incidents requiring acute or emergency care, 402 (89.1%) resulted in transportation to the hospital. Only 18 (4.5%) respondents had ever used an ambulance in the emergency situation (    Table 4.

Affordability
Of those respondents who knew the existence of a public-access number for medical emergencies, only 76 (37.1%) knew it was a toll-free call. About one-third (n = 167, 35.7%) of all the respondents indicated they would be more likely to call the 1-9-3 number in an emergency if they knew the call was toll free. When asked about their perceived cost of ambulances as a potential barrier, the majority of subjects indicated that government ambulances were free or affordable (n = 249, 53.2%) and that private ambulances were too expensive (n = 235, 50.2%).

Acceptability
The majority (n = 256, 54.7%) believed the EMTs offer high-quality care, and most (n = 403, 86.1%) believed it is overall ''better'' to go by ambulance in an emergency. However, the majority also thought taxis are faster than ambulances in Accra (n = 365, 78.0%).

Hypothetical Scenarios
Of the 459 total responses to both hypothetical questions, 57 (12.4%) respondents reported they would call for an ambulance (either NAS or other) in both hypothetical scenarios (witnessing a pedestrian severely struck by a vehicle and a family member being burned badly in a house fire) ( Table 5). The most common response was to use a taxi in both scenarios (n = 225, 49.0%) while 21 (4.6%) stated they would take any available vehicle as a means of emergency transportation in both events. An analytic model was developed in which the likelihood of calling an ambulance after witnessing a pedestrian-auto collision (outcome variable) was assessed as a function of respondent demographics of age, sex, education level, cell phone ownership, car ownership, prior ambulance experience, knowledge of ambulance function, knowledge of cost of the 1-9-3 call, and perception of ambulances (safe, fast, better). Age of respondent (in years) was found to be negatively associated with the likelihood of calling for an ambulance (Table 6). This response was then categorized into 3 age variables: 18-35, 36-50, and over 51 years. Being between the ages of 18 and 35 (P = .01) and believing ambulances were safe means of emergency transport (P = .02) were significantly associated with reported likelihood of respondents to call for an ambulance in the case of an emergency. Prior experience with an ambulance (P = .06) was positively associated with likelihood of calling for an ambulance, but the association was not statistically significant. Those respondents aged 18-35 were 2.3 times as likely to call for an ambulance (odds ratio [OR], 2.27) as older respondents, and those who believed an ambulance is safer than a taxi were over 2 times as likely to report they would call for an ambulance (OR, 2.17) as those who did not hold such beliefs, while those who had personal experience with an ambulance were 75% more likely to report they would call for an ambulance than those with no prior experience 69% of respondents thought the number of ambulances in Accra insufficient.
Of those respondents who knew about the public access emergency number, only 37% knew it was toll free.
In hypothetical emergency scenarios, respondents most commonly reported taxis as the preferred transportation mode.
(OR, 1.75). No other statistically significant association was discovered between any other variable and likelihood of calling an ambulance in the pedestrian injury scenario.
A similar regression was repeated using the second hypothetical scenario (in which a family member was burned in a house fire), with the respondents' answers to how they would transport that patient to the hospital as the outcome variable, while maintaining the original independent variables (listed above). Results were similar (data not presented).

DISCUSSION
Although the significant burden of acute disease in Ghana suggests a large need for ambulance services, calls for ambulances from the public are   disproportionately lower. 9,15,16,29 Our study has elucidated contributing factors to this issue in Accra, where study participants most commonly indicated they would call a taxi-perceived to be faster than ambulances-in hypothetical traumatic emergency situations. The majority of participants thought there was an insufficient number of ambulances. Yet most participants had favorable opinions of existing ambulances, indicating they provided high-quality care and that they were safer and ultimately better than taxis. Only a minority, however, knew of the toll-free public access number, and many thought that ambulances should be used to transport corpses. With awareness of proper use of ambulance services and how to access it, along with adequate numbers of ambulances per population, use of ambulance services in emergency medical situations may improve. Our main findings, interpretations, and their significance are discussed below in the context of respondents' prior experiences, preferences, and perceived barriers and facilitators.

Prior Emergency Medical Experiences and Use of Ambulance Services
Up to 75% of respondents reported experiencing a medical or traumatic emergency within the preceding 5 years. Among these, only about 5% had used an ambulance for emergency transportation to a health care facility. This is a lower proportion of ambulance use than was found in prior studies in Ghana that indicated ambulance use by the general population in 8% to 12% of emergencies. 9,11 This difference may be accounted for by respondent recall bias, although we do acknowledge that emergency situations are often major life events and less likely forgotten; however, we have no way to assess to what extent participants were accurately recalling events. In addition, this study excluded cases of interhospital ambulance transfers, which are relatively common in Ghana and were included in prior studies. 9,11,16 Nonetheless, the results confirm a generally low rate of ambulance access requests and utilization for medical and traumatic emergencies in a resource-constrained African setting.

Preferences for Modes of Emergency Transportation
When presented with the study's hypothetical emergency injury scenarios (pedestrian hit by a car and person burned in a house fire), relatively few (12%) Ghanaians stated they would call for an ambulance in both scenarios, while 49% said they would rather use a taxi in both scenarios. These findings may be explained by the fact that 75% of respondents perceived taxis to be a ''faster'' form of prehospital transport than ambulances. Although there is no prior research assessing taxi response times in Ghana, given the ubiquitous presence of taxis and relative paucity of ambulances in 2013, it seems reasonable to assume a faster response time from the former. 16 Paradoxically, respondents generally had favorable opinions of existing ambulance services in Accra: over half (55%) felt that EMTs offer ''high-quality'' care, 81% acknowledged that ambulances were in fact the ''safer'' form of emergency transport, and 86% affirmed that ambulances were ultimately the ''better'' way to be transported during emergencies. However, 69% of our sample felt the number of ambulances in the Accra metropolitan area insufficient. This leads to the inference that when an adequate number of ambulances exists, and citizens are made aware of this service, individuals (who appear to generally understand the appropriate indications to access an ambulance) may be inclined to actively seek access to them more often for primary, or scene, responses. Of note, given the high percentage of respondents who indicated some preference for taxis, a complementary, practical non-ambulance solution for safer emergency transport may include mass training of taxi drivers in first aid, appropriate destination hospital selection, and safe emergency transport (successfully demonstrated by Mock et al. in a small pilot study in Kumasi, Ghana, in the 1990s). 9,10 Perceived Barriers and Facilitators to Accessing Ambulances Although 44% of respondents who knew the existence of a public access number in Accra, only 37% knew it was toll free, and 36% stated they would be more likely to call if they knew the call was free. Half believed NAS (the government ambulance service) was affordable, but very few Believe ambulance safer than taxi 2.17 (1.12, 4.19) .02 Abbreviations: CI, confidence interval; OR, odds ratio.
When adequate numbers of ambulances exist and citizens are aware of the service, individuals may be more inclined to use ambulances.
(3%) knew it was a free service. Misperceptions about telephone access and cost of ambulance services are indicative of a general paucity of education among Ghanaians in Accra, and this is most likely generalizable to citizens across the nation. Issues within Ghana that focus on telephonebased access to emergency medical services through emergency medical dispatch are the subject of a parallel study by the investigators of this study. We demonstrated that those with prior personal experience with an ambulance service, who were younger in age, or who believed ambulances to be a safer means of transport were more likely to report they would call for an ambulance in the event of a medical emergency.
Several previously hypothesized barriers were not perceived by our participants as obstacles to accessing EMS services, including poor knowledge of ambulance companies and appropriate indications to access ambulances. Over 90% of respondents were aware of the existence of at least one ambulance company in Accra, and about half knew specifically of NAS. Over 90% of surveyed citizens knew at least one appropriate indication for accessing an ambulance. Interestingly, 23% of respondents also inappropriately suggested that ambulances in Ghana should be used to transport corpses, a long-standing accepted practice in Ghana that was outlawed in 2012. This warrants special mention as many respondents felt ambulances were generally vehicles for transporting corpses, a common misperception that may prevent appropriate ambulance access. This deep-rooted sociocultural perspective may persist until this historic practice is completely aborted in Ghana and purposeful public information reverses the misperception.
Prior to this study, only one published report in the African emergency care literature was found that sought to assess barriers to accessing prehospital emergency care by a medically undifferentiated population. 30 That study was undertaken in Libreville, Gabon, in 2009, where investigators conducted a brief 9-question oral interview of a small (N = 25) convenience sample of patients and visitors at a local emergency center. Qualitative results from this study indicated that misperceptions, lack of awareness, alternative forms of transport, and cost were all barriers to accessing prehospital resources. While our results support many of these findings, cost does not appear to be as substantial a barrier as in this previous study. Further, the majority of our study's population had high awareness of the service and confidence in the service's care and safety. However, the specifics about how to access the service, indicated by the low prevalence of awareness of the free public access phone number, reflect a population-based problem of access to care in Ghana.

Possible Interventions to Improve Use of Ambulance Services
If the culture of seeking formal emergency transportation and professional prehospital medical care does not shift among populations, EMS systems risk failing in their mission to save lives when timesensitive expertise-driven care is required. Fortunately, our findings bear favorable implications for public health education as a means to appropriately improve access to and use of ambulance services in Ghana, given the preponderance of short-term modifiable factors. For example, focused public education about the toll-free number, ambulance safety, and response or transport times, delivered via media consumed by persons aged 18-35, may be one highyield targeted method to improve appropriate public ambulance utilization, once adequate ambulance response units exist in the near future. This may be accomplished as part of public media campaigns and/or incorporated into school-based education. However, there has purposefully been no large-scale public educational efforts by the Ghana Ministry of Health or NAS given the small capacity of this ambulance service relative to its large jurisdictional responsibility.
It is noteworthy that while the development of an EMS system is important and prehospital care confers a survival benefit, there must be concurrent A NAS ambulance on standby at a busy marketplace in central Accra, Ghana.
About 1 in 5 respondents thought ambulances should transport corpses.
improvement in in-hospital emergency care to synergistically improve patient outcomes. Emergency medicine is a recognized medical specialty in Ghana, and while specialist training has been ongoing in Kumasi, it is yet to commence in Accra where there is a paucity of emergency medicine specialist physicians. 9,15 Adaptation of the Survey Instrument While this survey instrument was designed for application in Ghana, plans are underway to adapt it to other African settings in which EMS leaders, administrators, and researchers desire to scientifically understand and quantify barriers to access of their local emergency medical services in order to increase the public's appropriate use of EMS services.

Limitations
We do note heavier sampling in business and academic centers, which may have introduced a selection bias toward more-educated individuals, thereby making our sample less representative of the Accra population. Analysis of the Accra study population's demographics indicate a high similarity to the general Ghanaian population; however, our conclusions are less generalizable to other urban areas in Ghana and least generalizable to non-urban areas in Ghana. Further, recall bias and decisions based on hypothetical emergency situations do not necessarily correlate with actual practice, and this is an unavoidable shortcoming in a survey.

CONCLUSIONS
Across Africa, EMS systems are developing rapidly in an effort to address the large burden of growing acute injury and endemic disease. This article describes the first known successful development and application of a robust community-based survey instrument to quantify the demographic, perceptual, and experiential factors that may prevent an African subpopulation from accessing critical EMS resources. Findings from this novel study indicate generally favorable perceptions of public ambulance services in Accra. Once ambulance agencies are operationally poised to receive an increased call volume and caseload, public education should be targeted to address selected topics, for example, to improve awareness of the toll-free public access number and about the inappropriateness of transporting corpses, thereby positively influencing citizens' decisions to call for an ambulance. The present findings also support consieration of reconfiguring the public ambulance service to encompass trained taxi services as firstresponders, which has been previously pilot tested in Ghana with operational and educational success.
Future work will include deploying the survey in rural Ghanaian settings and other African countries developing EMS systems.