Study design
An observational study was designed which employed both quantitative and qualitative research methods. The study was conducted between May 2008 and February 2009 and consisted of three components: i) a retrospective audit of prescriptions to determine injection use patterns; ii) systematic observation of UHCs facilities, injection providers, waste handlers, and injection administering events to assess how injection safety measures were followed; and iii) six focus group discussions (FGDs) with doctors and 38 in-depth interviews (IDIs) with doctors, nurses, and waste handlers on injection safety practices. Qualitative data supplemented the quantitative findings for practices that were not identified in the survey.
Study sites
This study took place in 24 primary health care centers (known as Upazila Health Complexes) of six divisions in Bangladesh. Geographically Bangladesh is divided into six divisions (namely Dhaka, Chittagong, Khulna, Rajshahi, Sylhet and Barisal). We randomly selected four Upazila Health Complexes (UHCs) from each of these geographical divisions. The UHCs are government health facilities located at sub-district level and offer mostly primary (in limited areas provide secondary level care) health care. Each UHC serves an estimated population of 0.5 million in rural Bangladesh. We excluded district level (which provides secondary level care) and Medical College Hospitals (provides tertiary level care) from this study, as our focus of interest was to assess injection practices in primary health care settings.
Data collection tools
The questionnaire for prescription audit and injection safety observation tools were developed based on previous research tools [12, 22, 23], which have been modified in accordance with the World Health Organization (WHO)/Safe Injection Global Network (SIGN) recommended revised Injection Safety Assessment Tool [24].
In addition to socio-demographic characteristics, the survey instrument included information on injection prescribing pattern and safety practices. The indicators included; total number of prescribed medications per patient, total number of injections used per prescription, percentage of prescribed injections per prescription, injection use rate (percentage of patients who received one or more injection), percentage of injection used in certain tracer health conditions (e.g. diarrhea, skin disease, fever), percentage of injection prescribed in specific health conditions, frequency distribution of types of injections (e.g. antibiotics, intravenous fluids, pain killers/analgesics, anti-histamines) given per tracer condition.
As outlined in WHO/SIGN revised Injection Safety Assessment Tool [24], injection safety practices were observed under four sub-categories, such as indicators for: i) health facilities and general safety for heath workers, recipients and community (e.g. proportion of facilities where injection safety or waste management policy/guidelines were available for viewing, percentage of facilities with no overflowing or pierced sharps containers, percentage of facilities with no used sharps in an open container or open areas, proportion of facilities with safe final waste disposal methods); ii) injection providers' safety (e.g. percentage of providers: reported needle stick injuries in the last six months, who were trained in injection safety practices, who were immunized against hepatitis B); iii) patients/injection recipients safety (e.g. percentage of facilities with no used sharps in open areas, percentage of events providers maintained antiseptic procedure, percentage of events where a new syringe- needles were used every time, percentage of recipients who reported a adverse event, percentage of adverse events followed up after injection procedures); and iv) safety of waste handlers (e.g. percentage of waste handlers: have had access to and used 'heavy protective gloves', received formal training in healthcare waste management in the last two years, fully immunized for hepatitis B, and reported needle stick injury in the last six months).
To facilitate data collection for interviews and focus groups two separate topic guides were developed for IDIs and FGDs respectively. The guides covered a range of issues like providers' perspectives on injection uses; justifications of injection uses; causes of unsafe injection practices; risk perceptions; present knowledge and practices on injection safety (they included knowledge about risk of needle stick injury, diseases likely to be transmitted, individuals at risk etc.); syringes and needles disposal; types of syringes available; monitoring and regulatory mechanism related to safe injection use; and barrier to safe injection practices.
The draft data collection tools for all components of the study were piloted in one health facility in Dhaka, which resulted in a number of modifications. The final version was written in English and then translated into Bengali. Finally, the Bengali version was back-translated into English to check for linguistic validity of the tools.
Definitions
For the purpose of this study we considered injections as any medications that were injected either intravenously or intramuscularly or subcutaneously; we also included intravenous (IV) fluid administration.
The injection use rate in a facility was defined as the percentage of prescriptions at a certain health facility that list at least one or more injections. It was expressed as: Number of prescriptions examined during the study period containing at least one or more injections/Total number of prescriptions examined in a certain facility × 100. The injection use rate per prescription was defined as the percentage of injection prescribed per patient's prescription.
Safe/unsafe injection: An injection was considered safe if it did not harm the recipient, did not expose the provider to any avoidable risks and did not result in waste that is dangerous for the providers, recipients and community [25, 4, 23, 24]. On the contrary, an injection was therefore considered unsafe if it harmed the recipient, exposed the provider to any avoidable risks and resulted in waste that was dangerous for the providers, recipients and community.
Injection safety was defined as practices that intended to prevent transmission of infectious diseases between one patient and another, or between a patient and healthcare provider, and also to prevent harms such as needle-stick injuries, and to ensure safe environment for providers, patients and community through appropriate management of dangerous medical waste [23, 24].
Sampling and data collection
A research team consisting of eight trained research assistants (postgraduate pharmacy and medical students) collected survey and observation data of the study. Monitoring and supervision was provided by the principal investigators (AKAC and CA) and other members of the investigation team (co-investigators). The interviews and focus groups data were collected by co-investigators with supports from the research assistants.
Gaining access to the research sites
In order to gain access to the research sites, we acquired written permission from the Director General of Health Services as per the requirement of the Ministry of Health, Bangladesh. At local level, further discussions were held between the research team and the local hospital administrators of each site in order to facilitate data collection process. Based on our initial discussions, the hospital authorities of each site organized group information sessions with the representatives from all categories of service providers, where we informed them about the research, and the nature and process of data collection. A written information sheet outlining the nature of the research and the process of data collection was also circulated during information sessions. During specific types of data collection, for example systematic observations, providers who were being observed were informed about the research and the nature of the observations on a one-on-one basis and finally informed written consent was taken from each of them.
Study component I: retrospective audit of prescriptions on injection use
We collected retrospective prescribing data on injection use from hospital records. For sample size calculation, we assumed an alpha error of 0.05, precision of 5%, a design effect of 2 and considered prevalence of injection use as unknown. Considering these assumptions and including 15% of missing or incomplete data and the need to adjust for possible confounders, it was determined that a sample size of 708 from each division was required for the study. To round up and to cover further short-fall, we considered 720 prescriptions from each division, which yielded a total sample size of 4320 for the audit (Figure 1).
A two-stage, cluster-sampling method was used for the sample size calculation. From the whole country, six geographic divisions were selected with a probability proportional to population size. As illustrated in Figure 1, location clusters (24 UHCs) were chosen randomly at the first stage with equal probability with an aim to enroll 4 facilities from each division (4 × 6 = 24) on the basis of accessibility, non-ongoing interventions and maximum scattering across the country. A fixed number of prescriptions (n = 180) was then chosen from each of the selected clusters (each UHC) with equal probability based on available list of the patient records (prescribed between January 2006 and January 2008) and using random number tables.
Study component II: systematic observation of injection safety practices
Prospective data on injection safety practices were collected through observation visits to each of the selected facilities. The research team members observed each health facility for 4 hours per day over a period of seven days. The team observed the activities of 120 injection providers, 48 waste handlers and 24 hospital facilities to access how safety measures were followed and to look for evidence of injection practices. A total of 672 hours of observation (28 hours per facility) took place in a range of clinical areas characterized by frequent use of needles and other sharps, and waste disposal areas. Injection providers' and waste handlers' activities were observed on the spot based on their availability during the period of observation visits. We also observed a total of 480 injections being administered by 120 injection providers.
During each visit injection safety practices were observed in terms of injection preparation, administration of injection to the patients and disposal of syringe/medical waste in order to give a clear understanding of whether the health providers followed standard guidelines or not [22, 23].
Study component III: in-depth interviews and focus group discussions with a range of providers
This phase of the study involved in-depth interviews (IDIs) with the doctors, nurses and waste handlers, and FGDs with doctors of 24-UHCs to explore issues that were unanswered from the audit and/or observation data, and to supplement the findings of each other components. Potential participants for IDIs and FGDs were recruited either through convenience or network sampling techniques during audit or observation visit to the facilities. A total of 38 semi-structured interviews with doctors (n = 18), nurses/medical assistants (n = 14) and waste handlers (n = 6) of UHCs and six FGDs (with 5-7 doctors in each group, n = 43) were conducted in the local language. All in-depth interviews and focus groups were audio recoded.
Informed consent was requested from each participant and those who agreed were interviewed. The study protocol was reviewed and approved by the experts and review committees at the University of Dhaka, Bangladesh and the University of Nottingham, United Kingdom.
Data analysis
All statistical analyses were conducted using SPSS version 15 (SPSS Inc., Chicago). Calculations of frequencies, proportions, and χ2 tests were used to assess the significance of relationships between outcomes and explanatory indicators.
All qualitative data sources (interviews, focus groups, field observation notes) were assembled and interview tapes were translated and transcribed verbatim. After becoming familiar (that involves reading and re-reading of transcript texts, reviewing notes) with the content of texts, we identified a thematic framework that enabled us to recognize emerging themes or issues in the data set [26]. Extensive discussions were held between the members of researcher team to explore the key themes. On the basis of these discussions, we developed coding matrices for thematic analysis based on Ritchie and Spencer's framework approach [26] and entered data from all sources into these. As far as the approach is concerned, during the analysis stage we sifted, charted and sorted gathered data in accordance with key issues and themes. We then identified sections of the data that correspond to a particular theme and placed the specific pieces of indexed data in charts that consist of the headings and subheadings (drawn during the thematic framework). Finally, we mapped and interpreted the analysis of the key characteristics as laid out in the charts [26]. This rigorous process allowed us to identify key themes, explore discourses, and compare these across data and/or respondents, and to generate new ideas where appropriate. Results were then compared and discrepancies were discussed with the wider group, and concepts were further refined.
Finally, qualitative data was organized, and the central themes/findings were constantly compared and combined with results of quantitative data to see how findings complemented each other. The key findings from both data sets were then analyzed in relation to wider perspectives of injection uses and injection safety practices (locally and where applicable internationally).