Diagnose-Specific Antibiotic Prescribing Patterns at the Otorhinolaryngology Inpatient Departments of Two Private Healthcare Facilities in Central India: A Five Year Observational Study

Antibiotics are overprescribed in low-and-middle-income countries where the infection rate is high. The global paucity of standard treatment guidelines and reliable prescription data is a barrier to rationalise antibiotic use and combat antibiotic resistance. Moreover, there is a lack of diagnose-specific prescription data from high infection risk departments such as the otorhinolaryngology (ENT). The purpose of the study was to present diagnose-specific antibiotic prescribing patterns at ENT inpatient departments of a teaching hospital and a non-teaching hospital from Indian private healthcare sector. Data of all consecutive inpatients at the department (n=3527) were collected for five years (2008-2013). Analyses were conducted for inpatients aged >15 years (n=2909) using the World Health Organization’s methodologies. Patient records were divided into four diagnoses (indication) groups, i.e., surgical, non-surgical, chronic suppurative otitis media (CSOM) and others.


Abstract Background
Antibiotics are overprescribed in low-and-middle-income countries where the infection rate is high. The global paucity of standard treatment guidelines and reliable prescription data is a barrier to rationalise antibiotic use and combat antibiotic resistance. Moreover, there is a lack of diagnose-specific prescription data from high infection risk departments such as the otorhinolaryngology (ENT). The purpose of the study was to present diagnose-specific antibiotic prescribing patterns at ENT inpatient departments of a teaching hospital and a non-teaching hospital from Indian private healthcare sector.

Results
Of 2909 inpatients, 51% had surgical diagnoses. An average of 83% inpatients in the clean surgery group, 78% in viral infection and 75% in non-infectious groups were prescribed antibiotics. CSOM was the most common diagnosis at both settings (883/2909 inpatients), where nearly 90% inpatients were prescribed antibiotics.
Overall, third-generation cephalosporins and fluoroquinolones were most commonly prescribed.

Conclusions
This study highlights prescribing antibiotic to the unindicated diagnosis groups, i.e., clean surgeries, viral infections, and non-infectious diagnoses. Recommended single-prophylactic dose of antibiotic was not prescribed for the majority of cleancontaminated surgeries. Prolonged empirical prescribing and insignificant use of the microbiology laboratory was evident in both settings. A meticulous analysis of clean surgery group highlighted the universal applicability issue of available global guidelines.

Background
Antibiotics are overprescribed worldwide but the practice is more common in lowand middle-income countries (LMICs), where standard treatment guidelines are not available or implemented, and the infection rates are high (1,2). Antibiotic resistance is an inevitable consequence of antibiotic consumption that limits the treatment options, increases morbidity, treatment failures, and mortality (1,2). The World Health Organization (WHO) has recognised antibiotic resistance as a global public health threat and identified local prescribing surveillance studies as a crucial step to slow down the emergence of resistance (1,2). National surveillance networks for the prescriptions, including antibiotics, do not exist in most LMICs and neither in India mainly due to financial, technical and human resource constraints, and scarcity of computerized medical record system (3). In LMICs where both the infection risk and use of the antibiotics are high (1,2), the data collection can be initiated at smaller scales. These small-scale studies are useful to inform the decisions made and over time, can be scaled up to national surveillance. Based on an estimation of national pharmaceutical sales data of 71 countries in 2010, India was the largest antibiotic consumer for humans (4). Thereby, India is also a presumed hub of antibiotic-resistant bacteria.
Antibiotics are prescribed profusely at healthcare facilities. These places also provide suitable conditions for rapid emergence and spread of antibiotic-resistant bacteria (3,5,6). The few available surveillance studies from India primarily focus on public sector healthcare facilities (5,(7)(8)(9)(10). However, a majority of the Indian population seek healthcare at private sector facilities, where the poor implementation of globally available guidelines and lack of local prescribing guidelines is evident (5,8,11,12). Studies from high-income countries report high prescribing of antibiotics to the patients admitted to the otorhinolaryngology departments, commonly known as the department of ear, nose, and throat (ENT) (13,14). Therefore, surveillance studies focusing departments with a high risk of infections in the private sector, such as ENT department, is crucial. In India, the patterns of antibiotic prescribing for ENT diagnoses have only been published from outpatient departments and primary healthcare facilities (15)(16)(17) and not from the inpatient departments of higher healthcare levels.

Method
The aim of the study was to present and compare the antibiotic prescribing patterns corresponding to the ENT diagnoses (indications), and further probing to the surgical and non-surgical indications, at the ENT departments of two private sector, tertiary healthcare hospitals, a teaching hospital (TH), and a non-teaching hospital (NTH), in Ujjain district of Central India.

Study Design
A cross-sectional study was conducted for all patients admitted for five years, 2008-2013, in the ENT inpatient departments at the TH and the NTH. The data were analysed and presented at the department level, diagnosis groups and specific diagnosis level.

Study Settings and Sample
The two study hospitals are located in Ujjain district of Madhya Pradesh state in Central India. The study settings have been described in previous publications in detail (5). In brief, the study settings were a TH that is associated with a private medical college and an NTH. The TH is located in a rural area of Ujjain district and had 570 hospital beds at the time of the study. The healthcare policies of TH could be compared with the Indian public-sector healthcare facilities as it provided the medical care and medicines on a full charity basis to all patients visiting the hospitals. The NTH is situated in the city area of Ujjain (350 beds) where patients paid the subsidised cost for the medical consultations, prescribed medicines, and the diagnostics. Diagnostic facilities including microbiology laboratories were readily available at both study settings. Diagnose-specific standard treatment guidelines were not available at any of the study hospitals at the time of data collection.
The study population consisted of all consecutive patients admitted to the ENT departments of the two study hospitals. Data of a) the inpatients aged less than 15 years old, b) those that did not have an ENT related diagnosis or c) had incomplete patient records were excluded before analysis (n = 550, Figure 1). Patients who stayed for at least one night in the ENT wards were considered as inpatients and were included in the analysis.

Data Collection and Process of Analysis
The data collection has been explained in detail in previous publications (5,8). In short, the nurses were trained for manual data collection using a contextualised data collection tool. The tool included basic demographics, admission and discharge dates, description of prescribed antibiotics such as name, dose, duration and frequency, and the diagnoses. The data was entered in Epidata software 3.1 (http://www.epidata.dk/download.php) and Microsoft Excel® software. In the present study, we have used the definition of surgical prophylactic antibiotic treatment, as presented in the SIGN classification system, i.e., the use of antibiotics before, during or after a diagnostic, therapeutic, or surgical procedure to prevent infectious complications (22)  The prevalence of chronic suppurative otitis media (CSOM) is high in the West Pacific, South East Asia and particularly in India (24). Antibiotics are commonly prescribed to patients with CSOM (24). Therefore, the prescriptions to the CSOM patients might influence the overall results of surgery patients. Thus, all patients with CSOM were categorised separately in group C. Group D contained other indications that could not be categorized in the groups A, B, and C such as atrophic rhinitis. The prescriptions for group D were not analysed in detail.

Results
Of the 3527 patients admitted to the ENT departments, 2909 were included in the analysis ( Figure 1). Out of included inpatients, 2358 underwent surgical procedures.
Of 2358, 883 had CSOM (group C, table 2) and the rest were categorised in group A.
Group B, the non-surgical indications, consisted of 537 inpatients whereas 14 inpatients had other diagnoses (group D). In both hospitals, CSOM was most common among patients aged between 15 and 50 years, and cancers were most common among the patients aged more than 50 years.
Overall in group A, 85% were prescribed antibiotics. At diagnosis group level, antibiotic prescribing was higher specifically in the dirty/contaminated surgery (A1) and clean-contaminated surgery (A2) subgroups at the NTH than at the TH (96% each compared to 94% and 93%) while it was higher in the clean surgery (A3) at the TH than at the NTH (86% compared to 81%, table 2). A diagnose-specific analysis of tonsillectomy and adenoidectomy in subgroup A3 showed that at the NTH, 36 patients underwent any of these surgeries and 32 were prescribed antibiotics with a median treatment period of 3 days (range: 1-6 days). At the TH out of the 109 patients who underwent tonsillectomy or adenoidectomy, 106 patients were prescribed antibiotics with a median treatment period of 6 days (range: 1-21 days).
Moreover, a higher proportion of the prescriptions of subgroups A2 and A3 at the NTH adhered to the international prescribing guidelines (20-25%) than at the TH (10-11%). 22,23 Overall, at the NTH, the proportion of prescriptions made using brand names (94%) was higher when compared to that at the TH (55%).
CSOM (group C) was the most common diagnosis with 20% of total inpatients at the NTH and 33% at the TH (table 2). At both hospitals, group C had the most extended duration of hospital stay and antibiotic treatment. Both durations were longer at that TH with a median of 11 days (range-1-38 days) and four days (range: 1-32 days) at the NTH. At the TH, 95% of the inpatients with CSOM were prescribed antibiotics and 89% at the NTH. One patient with CSOM at the NTH was prescribed antibiotics ear drops.
In the non-surgery group (group B), antibiotics were prescribed to a higher proportion in all subgroups at the TH (83-94%) than at the NTH (75-92%, table 2).
In the non-infectious (subgroup B3), 75% were prescribed antibiotics at the NTH and 83% at the TH. The duration of antibiotic prescription was highest in groups A, and C. Figure 3 presents the duration of antibiotics prescription in days to the inpatients of the groups A and C at the subgroup level. Among the patients in subgroup A2 (cleancontaminated surgery), 96% at the NTH and 93% at the TH were prescribed antibiotics for a longer duration than the recommended duration for this group (24 hours).
At discharge, a significantly higher number of patients were prescribed antibiotics at the TH (43%) than at the NTH (33%, p<0.05). The patients in the surgical diagnosis groups (A and C) were more often prescribed antibiotics than the patients in non-surgical group B. Overall, the FDCs of amoxicillin with clavulanic acid J01CR50 (45%), cefuroxime J01DC02 (19%) and ciprofloxacin J01MA02 (8%) were the most commonly prescribed antibiotics at discharge.
Overall two deaths were reported at the TH, one with CSOM and the other with tuberculosis. No co-morbidities were reported on these patients. Overall, the samples of 0.5% inpatients were sent for bacterial culture and antibiotic susceptibility tests from both settings. In total, the antibiotic therapy was changed in 0.4% prescriptions based on the microbiology culture and susceptibility reports.

Discussion
To the best of our knowledge, this is the first long-term study at diagnosis level that describes the pattern of antibiotic prescribing among ENT inpatients of Indian hospitals. Due to lack of ENT department-specific studies from the LMICs, we have compared our results with the studies of other departments. The present study highlights that antibiotics were prescribed to almost nine of the ten admitted patients (TH-91% and NTH-86%). Antibiotics were prescribed most frequently and for the most prolonged period to the CSOM inpatients. Antibiotics were commonly prescribed for clean surgeries, clinically viral infections and non-infectious diagnoses. Overall, empiric treatment was extended during the entire hospital stay for most inpatients. One of the reasons for this extension might be paucity in sending samples for bacterial culture and antibiotic susceptibility tests. None of the inpatients was prescribed a single prophylactic dose of antibiotics, for clean-contaminated surgeries, as per recommendations (22,23). Ceftriaxone and ciprofloxacin were predominantly prescribed in all groups. In India, increasing prevalence of bacteria resistant to carbapenems have been observed (4).
Encouragingly, no carbapenem was prescribed at any of the ENT departments.
Prescribing patterns in surgical diagnoses group, Group A Among the patients who had clean-contaminated surgery, subgroup A2, a single dose regime of a first-or second-generation cephalosporin, preferably cefazolin, is recommended as prophylactic antibiotic treatment (22,23). In our study, a thirdgeneration cephalosporin was the preferred antibiotic choice in both settings.
Similar results were shown in two studies conducted at general surgery department in Western India (26) and at a private tertiary healthcare hospital in Southern India (27), where third-generation cephalosporins were reported as the most commonly prescribed class of antibiotics. The preference to a broader spectrum antibiotic, such as the third-generation cephalosporin, might be due to the overall increase in prevalence of bacterial resistance to antibiotics in India that has been observed in other settings (4).De-escalation of antibiotic therapy is suggested for settings where broad-spectrum antibiotics are the first drug of choice, but was not found at our settings (28). Overall, nine out of ten patients who underwent clean-contaminated surgery were prescribed prophylactic antibiotic treatment for more than one day. This might be due to a common global misunderstanding, as presented by Bratzler D et al., that longer prophylactic antibiotic treatment periods are considered to be more effective to prevent surgical site infections (SSI) than single dose regimes (23). Prescribing broad-spectrum antibiotics during a prolonged period as prophylactic antibiotic treatment is not recommended due to the increased risk of adverse effects, the risk of emergence of antibiotic resistance and higher treatment costs (1,6,23).Studies to advocate prescribing a single dose of a first-or secondgeneration cephalosporin and to develop and introduce local prescribing guidelines based on the susceptibility patterns are also suggested.
Among the patients who underwent clean surgery (subgroup A3), 81% at the NTH and 86% at the TH were prescribed antibiotics, even though prophylactic antibiotic treatment is not recommended for this subgroup (22,23). Khan

et al. from Southern
India, reports that many surgeons often take a 'safety approach' and prescribe antibiotics to minimise the risk of SSI in clean surgeries (27). It is worth mentioning that the risk of SSI in clean surgery is reported to be less than 2% (22). Reasons to use a 'safety approach' could be a high patient burden per physician or presumed poor hygienic conditions of the patients as the catchment area included villages of low socio-economic status (28).All these presumptions need to be verified through a suggested qualitative study.
The results of antibiotic prescribing practices for tonsillectomy and adenoidectomy in the subgroup A3 highlights several underlying issues. The first issue was in accordance with a globally ongoing discussion. The discussion raises the question of whether antibiotics should be prescribed as a prophylactic antibiotic treatment to the patients undergoing tonsillectomy or adenoidectomy or not (22,23,29). One side of the debate is presented by the USA and SIGN guidelines. These guidelines suggest to prescribed antibiotics as a general prophylactic antibiotic treatment for the above mentioned surgical procedures, but only to the patient's having risk factors of acquiring bacterial infections (22,23).The other side of the debate presents a regional scenario based on local infection risk factors, as mentioned in a textbook of ENT diseases written by Indian authors (29).According to the recommendations of the book, prophylactic antibiotic treatment can be prescribed post-operatively up to a week (29). The ENT physicians at our study settings might have followed the local suggestion of prescribing antibiotics for the surgeries in question, and this could also explain the overall high proportion of antibiotic prescriptions in the subgroup A3.
Another issue raised was regarding universal applicability of a guideline. In the present study, the American and Scottish guidelines were used to facilitate the classification of the surgical indications concerned with prescribing of antibiotics (22). On the other hand, the local prescribing rationale suggests classifying tonsillectomy and adenoidectomy in subgroup A1, i.e., dirty/contaminated surgery and not in A3, i.e., clean surgery (29). Moreover, the use of different surgical techniques and varied access to the resources in a setting also affect the risk of surgical site infections and are a cause of prescribing antibiotics (30). Hence, the present diagnoses-specific study highlights our concern of applicability of available Prescribing patterns to the inpatients with nonsurgical diagnoses, Group B Among the inpatients with non-surgical diagnoses, group B, antibiotic treatment is indicated only to the patients having a microbiologically confirmed or clinically estimated high risk of a bacterial infection (subgroup B1) (22,23). A majority of the patients in subgroup B1 of both hospitals were prescribed antibiotics as per the recommendations. However, antibiotics are not indicated for clinical infections of suspected viral origin for subgroup B2 or non-infectious diseases, and subgroup B3 (31). Still, 76% patients at the NTH and 84% patients at the TH in these subgroups were prescribed antibiotics. There is a need to conduct a separate probing study to explore the underlying factors for this practice.
Furthermore, a longer duration of hospital stay has been reported to correspond to a higher risk of acquiring healthcare-associated infections (HAI) and increases the risk of being prescribed antibiotics (32). Thus, in the present study, a preventive approach for HAI might be the reason for the higher proportion of patients being prescribed antibiotics at the TH compared to the NTH. However, the DDD/100 prescriptions were higher at the NTH than at the TH, pointing towards prescribing high antibiotic doses per prescription at the NTH. Prescribing antibiotics for nonindicated conditions, higher doses and for longer periods than indicated are considered as preventable factors that, if continued, might accelerate the increase in antibiotic resistance (1,2,6).
Prescribing patterns to the CSOM inpatients, Group C CSOM was the most common diagnosis in our study settings. The procedures related to CSOM are categorised in classes comparable to subgroup A1 (dirty/contaminated surgery, e.g., emergency mastoidectomy), to which antibiotics should be prescribed as treatment and A3 (clean surgery, e.g., tympanoplasty), to which antibiotics should not be prescribed (22,23,33). In group C, antibiotics were prescribed to 95% inpatients at the TH (NTH-89%), and 91% were prescribed antibiotics for more than 48 hours (NTH-74%). The difference between the hospitals could be due to the presence of a higher number of contaminated/dirty surgeries performed at the TH than at the NTH. Furthermore, a WHO review article on CSOM, reports that topical antibiotics are superior to systemic antibiotics in terms of efficacy and have an advantage of less contribution to the development of antibiotic resistance (24). However, only one patient was prescribed antibiotic ear drops. The patients in group C at the TH had the most prolonged period of prescribed antibiotics with a median of 11 days. The lack of comparable antibiotic surveillance studies at ENT inpatient departments from the South-Asian region, lack of qualitative studies and consequently, the lack of antibiotics prescribing guidelines for CSOM surgeries restrict the possibilities to describe the reasons for this pattern antibiotic prescribing in the settings.
The new FDCs of antibiotics were prescribed at a lower extent (J01RA, Figure 2) as compared to the studies from departments at the study settings (5,34). Most of the new FDCs of antibiotics have no underlying scientific justification and does not add to drug's efficacy; however, adds to the cost of therapy, increase adverse effects and encourages antibiotic resistance (35).Thus, less prescribing of new FDCs in both settings could be appreciated.
The low practice of sending samples for bacterial culture and antibiotic susceptibility tests at both settings resulted in extended empirical prescribing throughout the hospital stay. This is similar to a short term study conducted in the settings (5). Empiric treatment is used to start antibiotic therapy for a suspected bacterial infection after collection of a sample from the suspected infection site for susceptibility and culture tests (6). The empirical prescribing ought to be reassessed based on the microbiological results (6). The active CSOM infections are often polymicrobial, and the patients have in most cases received several antibiotic prescriptions at lower healthcare levels before visiting and being admitted at the tertiary level healthcare settings (24,33). The high risk of multiple antibiotic treatments before hospital admission, suggests a low threshold for sending samples before initiating an empiric antibiotic therapy at the study settings, which was not observed (6). Results of our study highlight the need to conduct similar surveillance studies at other settings followed by contextualised qualitative studies to understand the underlying factors affecting the practitioners of the observed antibiotic prescribing patterns.
The strengths of this study were the long-term, continuous data collection, and detailed data at a diagnosis level recorded for all admitted patients. Furthermore, the data were from the private sector that provides healthcare to most of the Indian population. A limitation of this study was that perioperative notes were not included, which could have facilitated the interpretation of the results. However, this was not the objective of the present study. Although the data collection process was supervised robustly, the possibility of human error during data collection and entry cannot be denied. However, the possible effect of such human errors was expected to be minor due to the large study population, long study duration, robust monitoring, trained data recorders, and data entry staff. The establishment of the manual process of data collection, coding of antibiotics and indications and data entry explains why the data is analysed and published at the current time.

Conclusions
The present study offers a unique insight into the otorhinolaryngology inpatient departments of the private healthcare sector in India and identifies several underfocused issues that need to be addressed in the future. In the present study, a high proportion of inpatients with non-indicated diagnoses, such as clean surgeries, viral infections and non-infectious indications were prescribed antibiotics. A deviation in diagnose-specific prescribing patterns, concerning the recommendations, indicates the issue of the general applicability of the guidelines prepared in the HIC in the LMICs. This highlights a need for the development of contextualised standard treatment guidelines based on local disease and prescribing data. Prescribing broad-spectrum antibiotics empirically and inadequate use of the microbiology laboratory was also observed in both hospitals.
The scarcity of surveillance studies of antibiotic prescribing is a limitation to develop prescribing guidelines in the LMICs. Similar surveillance studies are explicitly needed from the private healthcare settings of the LMICs. Our sustainable, low-cost solution of manual data collection could be adapted at the resourceconstrained settings to reduce delays in conducting surveillance studies due to lack of computerization. Furthermore, the use of ATC/ DDD methodology and the ICD-10 codes make our results comparable and exchangeable to the other global studies.
The findings of this study to formulate a standard policy and practice of the patterns of antibiotic prescribing in high infection risk inpatient's departments.
The issues raised in present communication need further probing using a qualitative approach for better understanding of the influencing factors for present prescribing patterns and possible areas of improvement. We also suggest developing

Consent for publication
Not applicable

Availability of data and material
The datasets generated during the current study are not publicly available due to breaching individual privacy but deidentifiacted datasets are available from the corresponding author on reasonable request. A summary of data analysed during this study are included in this published article and its supplementary information files.  .Selection of prescribed antibiotics at diagnosis subgroup level using the DU90 approach. 10 Duration of antibiotic treatment to the patients of Group A and C in the study hospitals. Abbr

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download. 190607_TableS1_BMC.docx