Antibiotic Prescribing Practices and Errors among Hospitalized Pediatric Patients Suffering from Acute Respiratory Tract Infections: A Multicenter, Cross-Sectional Study in Pakistan

Background and objective: The noncompliance of treatment guidelines by healthcare professionals, along with physiological variations, makes the pediatric population more prone to antibiotic prescribing errors. The present study aims to evaluate the prescribing practices and errors of the most frequently prescribed antibiotics among pediatric patients suffering from acute respiratory tract infections who had different lengths of stay (LOS) in public hospitals. Methods: A retrospective, cross-sectional study was conducted in five tertiary-care public hospitals of Lahore, Pakistan, between 1 January 2017 and 30 June 2017. The study population consisted of pediatric inpatients aged 0 to 9 years. Results: Among the 11,892 pediatric inpatients, 82.8% were suffering from lower acute respiratory tract infections and had long LOS (53.1%) in hospital. Penicillins (52.4%), cephalosporins (16.8%), and macrolides (8.9%) were the most frequently prescribed antibiotics. Overall, 40.8% of the cases had antibiotic prescribing errors related to wrong dose (19.9%), wrong frequency (18.9%), and duplicate therapy (18.1%). Most of these errors were found in the records of patients who had long LOS in hospital (53.1%). Logistic regression analysis revealed that the odds of prescribing errors were lower in female patients (OR = 0.6, 95% CI = 0.1–0.9, p-value = 0.012). Patients who were prescribed with ≥3 antibiotics per prescription (OR = 1.724, 95% CI = 1.1–2.1, p-value = 0.020), had long LOS (OR = 12.5, 95% CI = 10.1–17.6, p-value < 0.001), and were suffering from upper respiratory tract infections (URTI) (OR = 2.8, 95% CI = 1.7–3.9, p-value < 0.001) were more likely to experience prescribing errors. Conclusion: Antibiotics were commonly prescribed to patients who had long LOS. Prescribing errors (wrong dose, wrong frequency, and duplicate therapy) were commonly found in cases of lower respiratory tract infections (LRTIs), especially among those who had prolonged stay in hospital.


Introduction
Acute respiratory tract infections (ARTIs), i.e., upper respiratory tract infections (URTIs) and lower respiratory tract infections (LRTIs), are responsible for increased morbidity and mortality rates among children [1,2]. Children are more prone to ARTIs because of their weak immune system and constant contact with persons who could be carriers [3]. According to an estimation made by the World The records of 1730 participants who did not meet the inclusion criteria were excluded from the study. Consequently, records of 11,852 inpatients were finally selected for the study (Supplementary S2).

Data Collection
A data collection form was developed consisting of four main parts: (1) characteristics of the patients, (2) diagnosis, (3) recommended antibiotics, and (4) prescribing errors (Supplementary S3). The 2nd edition of the International Classification of Primary Care (ICPC-2) was used to define ARTIs as a clinical diagnosis of LRTIs, acute bronchitis (R 78), pneumonia (R 81), and epiglottitis (R 83). The URTIs included tonsillitis (R 76), whooping cough (R 71), acute bacterial sinusitis (R 75), laryngitis and tracheitis (R 77), and other URTIs, including common cold (R 74) [20]. The Anatomical Therapeutic Chemical (ATC) classification system [21] was used for the coding of antibiotics. The data relating to dose, dosage form, frequency, and duration of prescribed antibiotics were extracted by reviewing patient records according to the criteria (Supplementary S4). Appropriateness of prescribing antibiotics was assessed by critically observing and evaluating each of following indicators: drug selection, dosage form, frequency, prescribed quantity, and duration of therapy (Supplementary S4). The assessment of antibiotic prescribing errors was made by taking into account the clinical judgments, nature and severity of underlying disease, and social factors that may contribute to lengthening the patient's stay time in hospital. Four investigators were assigned to each selected setting; the principal investigator trained all investigators prior to the survey for the collection and validation of data. During the survey, two investigators filled out the investigational form, and the other two reviewed the medical records of patients. The LOS was evaluated by measuring the difference between the date of admission and the date of discharge [22].

Definitions
Prescription errors: Incorrect or inappropriate drug selection (based upon indications, contraindications, and other factors), dose, route, rate of administration, frequency, or duplication of drugs. Wrong frequency: Drugs prescribed in a frequency other than that recommended by the guidelines. Wrong dose: Prescribed dose greater than or less than 10% of the recommended dose in the guidelines. Wrong route: Prescribed drug administered by a route other than the recommended route mentioned in the guidelines. Wrong duration: Drugs prescribed for a duration other than that recommended in the guidelines. Duplicate therapy: Two or more drugs prescribed from the same class.

Data Analysis
Descriptive statistics were used to present the data. Furthermore, logistic regression analysis was performed to figure out the factors associated with prescribing errors. Results were expressed as odds ratio (OR) accompanied by 95% confidence intervals (95% CI), and a p-value < 0.05 was used for statistical significance of differences. Statistical Package for Social Sciences (IBM Corp.

Ethics Approval
The ethical approval was obtained from the Pharmacy Research Ethics Committee (PREC) at Akhtar Saeed College of Pharmaceutical Sciences (Reference: 19-2016/PREC, 26 December 2016). Before conducting the study, permission was obtained from hospital administrators.

Antibiotic Prescribing Practices
As the present study was conducted in multiple settings, different international guidelines were found to be implemented for different purposes. It was observed that the Current Medical Diagnosis and Treatment (CMDT) [23] and WHO guidelines were used for diagnosis, while the British National Formulary (BNF) [24] was consulted for doses. The American Academy of Family Physicians (AAFP), the American Academy of Pediatrics, the National Institute of Health and Clinical Excellence (NICE) guidelines on RTIs [25], and the American Thoracic Society Consensus Guidelines were consulted for treatment of pneumonia [26]. The IDSP guidelines [27] were considered for assessing the rational use of antibiotics.
A total of 26,883 antibiotics were prescribed to 11,852 patients. Out of this, 86.9% (n = 23,383) of the antibiotics were prescribed to patients suffering from LRTIs, while 13.1% (n = 3500) were prescribed to patients with URTIs. The practice of prescribing antibiotics was found to be more frequent among patients who had long LOS (52.9%, n = 14,217) compared to those who had short LOS (47.1%, n = 12,666). Most of the children were hospitalized due to LRTIs, especially pneumonia (short LOS = 72.9%, long LOS = 90.8%) (Supplementary S5 and S6). Amoxicillin (short LOS = 82.7%, long LOS = 82.8%) and ampicillin (short LOS = 61.8%, long LOS = 61.7%) were the most commonly prescribed antibiotics in pneumonia. Tonsillitis and pharyngitis were the most commonly observed URTIs among inpatients, and azithromycin (short LOS = 31%, long LOS = 47.9%) and amoxicillin (short LOS = 48.1%, long LOS = 30.8%) were the most frequently prescribed antibiotics for this condition (Supplementary S5 and S6).
In ARTIs, antibiotics were more frequently prescribed to patients who had long LOS in hospital compared to those who had short LOS. Penicillins (52.4%, n = 14,096), cephalosporins (16.8%, n = 4506), and macrolides (8.9%, n = 2404) were among the most commonly prescribed classes of antibiotics (Supplementary S7). Amoxicillin, ampicillin, ceftriaxone, and azithromycin were commonly prescribed antibiotics to patients who had long LOS compared to those who had short LOS in the healthcare settings (Table 2).

Antibiotic Prescribing Errors
A total of 4836 (40.8%) records had prescribing errors. Most of the errors were found among patients with LRTIs (82.8%) compared to URTIs (17.2%). 47.2% cases had one error, 21.7% had two errors, while more than three errors were found in 30.9% of those cases in which errors were identified. Wrong dose (especially underdosing), wrong frequency (especially high frequency), and duplicate therapy were the most common errors (Table 3). A large number of errors were found in the records of patients who were hospitalized for long periods of time (45.8%) compared to those who had short LOS (35.2%) (Supplementary S8).

Determinants Associated with Prescribing Errors among Pediatrics Patients
Logistic regression analysis revealed that the odds of prescribing errors decreased by 33% in female patients (OR = 0.6, 95% CI = 0.1-0.9, p-value = 0.012) compared to male patients. The odds of prescribing errors were 1.7 times higher in patients who were prescribed with ≥3 antibiotics per prescription (OR = 1.7, 95% CI = 1.1-2.1, p-value = 0.020) compared to those who were prescribed with one antibiotic per prescription. In patients with long LOS, the odds of prescribing errors were 12.5 times higher (OR = 12.5, 95% CI = 10.1-17.6, p-value < 0.001) compared to patients who had short LOS, while the odds of prescribing errors in patients suffering from URTIs were 2.8 times higher (OR = 2.8, 95% CI = 1.7-3.9, p-value < 0.001) compared to those who were suffering from LRTIs (Table 4).

Discussion
The present study set out to determine the antibiotic prescribing practices among the pediatric population suffering from ARTIs and the medication errors in the prescription records of inpatients with different LOS in hospital.

Antibiotic Prescribing Practices among Pediatric Patients
Our findings revealed that most of the patients suffering from LRTIs were prescribed with penicillins, while macrolides and cephalosporins were prescribed for URTIs. Among these, amoxicillin, ampicillin, and gentamicin were more frequently prescribed to patients who had long LOS compared to those who had short LOS in the healthcare settings. Evidence from previously published studies suggests that a majority of pediatric patients living in Pakistan are prescribed with cephalosporins after every five hours [28,29]. The unnecessary antibiotic prescribing practices in childhood ARTIs of viral origin is responsible for spreading resistant microbial strains and thereby increasing the disease burden [29]. These findings are comparable with a Jordanian study in which penicillins and cephalosporins were prescribed to a large proportion of the pediatric population [30]. In contrast to developing countries, the rate of antibiotic prescribing among ARTI patients declined from 5.2 to 2.2 prescription per 1000 pediatric patients during the last decade in various states of the UK [31]. In Pakistan, a survey-based study at a national level revealed that ARTIs were the most common indications for prescribing antibiotics and that, in most of the cases, the prescribing practices of many healthcare professionals were irrational [32].
Like the previously published literature [33][34][35], the present study also revealed that antibiotics were more frequently prescribed to patients who had long LOS in hospital compared to those who had short LOS. Most of the ARTIs (such as bronchitis) are of viral origin, and the IDSP guidelines recommend limited use of antibiotics. However, prescribers are compelled to prescribe antibiotics because of their inability to differentiate between viral and bacterial infections; parental demand to prescribe antibiotics; socioeconomic pressures; and to reduce the risk of secondary bacterial infections, duration of cough, and other symptoms [36][37][38]. The noncompliance of these guidelines was also observed in the present study. Macrolides were frequently prescribed in bronchitis, which can be managed through symptomatic treatment. In cases of pneumonia, ampicillin was prescribed as 57 mg/kg/day three times a day instead of 100 mg/kg/day four times a day [39]. These irrational practices are also attributed to the fact that, in Pakistan, pharmacists have a major role in procurement and dispensing of medicines but have limited exposure to the clinical setup and direct patient care. To implement rational antibiotic prescribing practices, the establishment of an antibiotic prescribing surveillance system and continual medical education programs need to be immediately implemented. In case of the pediatric population, clinical pharmacists must participate in parental counseling training programs of healthcare professionals.

Antibiotic Prescribing Errors
The present study revealed that most of the prescribing errors were related to wrong dose (especially underdosing), wrong frequency (especially prescribed greater than the recommended frequency), and duplicate therapy, particularly in the records of patients who had long LOS compared to those who had short LOS. These findings are in line with a retrospective study conducted on Brazilian pediatric patients in which long LOS was found in patients who had been prescribed with antibiotics [40]. Similarly, a prospective study depicted that higher rate of dosing errors in the medication record can threaten patient safety, especially in the pediatric population, because dose calculation is often more complex for these patients [41]. Wrong dose and frequency errors were commonly found among pediatric patients because many of the antibiotics were not readily available in pediatric unit doses, and pediatric doses were therefore calculated from adult unit doses. Antibiotic prescribing errors that are related to their frequent use may not only economically burden the healthcare system by increasing the patient's LOS in hospital but also subsequently play a crucial role in spreading antibiotic-resistant microbial strains [42]. Nonadherence to national and international guidelines and national formularies for the pediatric population could be the possible causes of these errors [15]. Therefore, hospital formularies and international standard treatment guidelines for pediatrics must be made available in all healthcare settings. To overcome the irrational prescribing of antibiotics, strict diagnostic criteria must be established for ARTIs of bacterial origin. The government must ensure availability of adequate number of hospital and clinical pharmacists in all pediatric public healthcare settings. Their prime responsibilities must include prescription evaluation, calculation of doses, reporting of antibiotic prescribing errors, and counseling of parents about the proper use of antibiotics. Moreover, the Pakistan Pediatric Association (PPA) must take initiatives to develop and implement Pakistan's own national formulary for pediatrics so that countrywide uniform treatment protocols are established. The inadequate healthcare system for pediatric patients has led to an increased incidence of prescribing errors and child mortalities. Low state spending on child health, abject poverty, and low literacy rate of parents has indirectly worsened the healthcare system in Pakistan [43]. Therefore, it is recommended that medical and paramedical staff members, along with senior consultants, are increased with the increase in patient load so that each patient gets adequate consultation time, and the risk of prescribing errors are reduced.

Factors Associated with Prescribing Errors
The prescribing errors were less likely to be found in the record of female patients. These findings are in line with previously published studies conducted in different states of Saudi Arabia and Pakistan, where pediatric male gender showed a greater association with prescribing errors [17,44]. Contrary to this, a meta-analysis showed that females were more susceptible to encountering prescribing errors than males [45]. Pakistan is an economically less developed country, where no public child healthcare insurance policy is available. Parents have to bear the health expense of their children on their own, and financial constraints and social norms compel most of the low-income households to prefer medical treatment for male children compared to female children. Due to this reason, male pediatric patients are more likely to be admitted in inpatient settings and are thus at higher risk of encountering prescribing errors than females [46].
Similar to previously published studies [47][48][49], LOS was found to be a significant factor for prescribing error. This might be due to the fact patients were prescribed with more number of antibiotics during their prolonged stay in the inpatient departments, which increased the risk of encountering antibiotic prescribing errors. Prescribing errors were more likely to be found in the records of patients suffering from URTIs compared to sufferers of LRTIs. These findings are consistent with previously published studies, where prescribing errors were less commonly found in the prescription records of patients suffering from LRTIs [50][51][52]. This may be attributed to the fact that LRTIs (like pneumonia) are more prevalent in the pediatric population than any URTIs, so specialized medical care is being provided to patients suffering from LRTIs. The prime focus of the available guidelines (such as the IDSP guidelines) is the management of LRTIs because, in most of cases, URTIs are usually mild and self-limiting [53]. Furthermore, statistical analysis revealed significant association between >3 prescribed antibiotics and higher risk of prescribing errors. These findings are consistent with previously published studies, which have shown that the number of antibiotics on prescription is a strong predictor of prescribing error in such a way that the risk of error becomes several folds with the addition of a single antibiotics in prescription [54][55][56]. This may be attributed to various reasons, including diagnostic uncertainties reflecting the physician's inability to distinguish between viral and bacterial infections, noncompliance of guidelines, poor practices of antimicrobial stewardship (AMS) program, heavy patient load, and unreasonable patient demands on physicians to prescribe antibiotics [57][58][59].

Strengths and limitations
To the best of our knowledge, this is the first study that deals with antibiotic prescribing errors among pediatric inpatients suffering from ARTIs. These findings will be beneficial to stakeholders in establishing policies and procedures for prescribing antibiotics to children aged 0−9 years. This study emphasized the need for an antibiotic prescribing errors reporting system, a national survey demonstrating rational use of antibiotics among the pediatric population, and the development and implementation of Pakistan own national formulary for pediatrics.
As with any study, this study also has limitations. First, the results cannot be generalized for Pakistan as a whole because clustering of data at different levels was not taken into account, and multilevel modeling was also not performed to test associated factors. Therefore, antibiotic prescribing practices may vary due to geographic variation of the selected healthcare settings, the prescribing habits of physicians, and the clustering effect. However, the healthcare sector conditions are similar across many regions of the country, and physicians consequently follow the same prescribing practices. We suspect that the findings and the issues raised will be similar in "like" hospitals (i.e., tertiary-care hospitals), so there are relevant learnings for the tertiary sector in Pakistan. Second, as this was a cross-sectional study, it was not possible to see the outcomes of antibiotic prescribing errors, i.e., adverse events and mortalities. Third, the findings of the present study are limited to pediatric patients suffering from only ARTIs because the inclusion of other ARTIs would have complicated the findings. Clinically based longitudinal studies may help to overcome these limitations.

Conclusions
The present study concluded that penicillins, macrolides, and cephalosporins were frequently prescribed classes of antibiotics in ARTIs, especially in the prescription records of patients who had long LOS. Among these, ampicillin, amoxicillin, azithromycin, and ceftriaxone were the most commonly prescribed agents. Prescription errors like underdosing, high frequency, and duplicate therapy were most commonly found among records of patients who suffered from LRTIs and were admitted for long periods of time in the public hospitals of Lahore. Caution must be exercised in prescribing antibiotics to the pediatric population because antibiotic prescribing errors can be hazardous to lives.