This was a retrospective cross sectional study which was carried out among inpatients of a tertiary care hospital in Kerala for two months (August-September 2020). The study was conducted after obtaining approval from the Institutional Ethics Committee (IEC Reg No: ECR/1299/Inst/KL/2019). Written informed consent was taken from each participant before accessing their medical records and the prevailing national ethical guidelines were followed [14].

Inclusion criteria: Inpatients receiving ≥3 drugs with duration of hospital stay of at least ≥3 days were included in the study.

Exclusion criteria: Patients with known drug allergies or history of drug abuse and emergency/intensive care unit patients were excluded from the study.

Sample size was calculated as 350 using the formula, n=z² pq/d², with 5% margin of error on either sides. Reference values for sample size calculation were obtained from the study done by Soherwardi S et al., among inpatients of tertiary care hospital [1].

Patients visited by the investigator were from inpatient locations like pay wards and general wards. Patients willing to take part were included as per inclusion criteria. Details required from each patient were name, unique identification number, age, sex, and diagnosis, details of drugs with dose, frequency and time of administration. All the relevant details collected were documented in data collection form. A drug interaction check was performed by the investigator using the Lexicomp drug interaction checker software powered by leading medical information suppliers Wolters Kluwer. The mechanisms of interaction can be subdivided into pharmacokinetic and pharmacodynamic. Pharmacokinetic interactions are those which can affect the processes by which drugs are absorbed, distributed, metabolised and excreted. Pharmacodyanamic interactions are those where the effects of one drug are changed by the presence of another drug at its site of action [4]. Prescriptions were reviewed again by the investigator before discharge of the patient, for any changes in therapy made by treating doctor to minimise the impact of DDI. Variables were coded and entered in Microsoft excel worksheet.

Percentage of prescriptions with DDI, total number of DDIs and average number of DDI per prescription were calculated as indicators of frequency. Mechanism behind each DDI were identified and documented. Drug interactions were categorised into major, moderate or mild based on severity. Major drug groups involved in interactions and the diseases with potential for DDI were identified. The changes in therapy advised by the doctor either by his/her own review or after receiving notifications regarding DDI were analysed and categorised.

Data were entered into Microsoft excel sheet and were analysed using Statistical Package for Social Sciences (SPSS) version 16. Descriptive statistics such as frequency, percentage, mean and Standard deviations were used. Chi-square test and independent sample t-test were used to find the statistical significance. A p-value of <0.05 was taken as statistically significant. Relation between duration of hospital stay and number of DDI was studied with Pearson correlation test.

I. Frequency of Drug-Drug Interactions (DDI): Total 350 prescriptions were evaluated during the study. Total number of interactions were 974 with an average of 2.78 DDI per prescription. Among the 350 prescriptions, 261 (74.6%) had at least one DDI. Results are tabulated in [Table/Fig-1].

II. Age wise distribution of Drug-Drug Interactions (DDI): Age group most prone for DDIs was 61-80 years. There was a significant difference between average number of DDI among patients ≤60 years and patients >60 years. Results are detailed in [Table/Fig-2].

III. Gender wise distribution of Drug-Drug Interactions (DDI): Out of 350 patients included in the study there were 156 males and 194 females. There was no significant difference between males and females regarding distribution of DDIs. Results are detailed in [Table/Fig-3].

IV. Relationship between number of drugs and number of DDI: Number of drugs per prescription ranged from 3 to 20 with an average of 8.90. Prescriptions were categorised into three groups according to the number of drugs. Maximum number of prescriptions was in the group 6-10. Percentage of prescriptions with DDI and average number of DDI were maximum in the group >10. Details are given in [Table/Fig-4].

V. Mechanisms and Severity of Drug-Drug Interactions (DDI): Out of the 974 interactions studied 306 (31.4%) were through pharmacokinetic mechanisms and 668 (68.6%) were through pharmacodynamic mechanisms. Most number of interactions was of moderate severity, followed by mild category. There is a significant difference in the frequency of major, moderate and mild interactions between pharmacokinetic and pharmacodynamic groups. Pharmacodynamic mechanisms were behind 72.7% of major and 73.7% of moderate interactions. Details are given in [Table/Fig-5].

Drug groups most commonly involved in interactions were antiplatelets, oral hypoglycaemic agents, bronchodilators, antibiotics, diuretics, insulin, statins, beta blockers, PPIs and NSAIDs. Details are given in [Table/Fig-6].

Diseases most commonly associated with interactions were hypertension, diabetes mellitus, dyslipidaemia, coronary artery disease, cerebrovascular accidents, respiratory tract infections and chronic kidney disease. Details are given in [Table/Fig-7].

One or more interventions were taken by doctor for minimising the impact DDIs. Most commonly taken steps were changing the timing of drug administration, monitoring for symptoms/signs/lab values/drug levels or both. Details are provided in [Table/Fig-8].

Analysis of the relation between duration of hospital stay and number of DDI showed a significant positive correlation with Pearson correlation value 0.406. Details are provided in [Table/Fig-9,10].

DDIs could be identified in 74.6% prescriptions collected for the study and the average number of interactions was 2.78. This is high when compared to the values observed by Soherwardi S et al., (66%) and Bollu M (57.44%) in tertiary care hospitals [1,4]. But in the study done by Kulkarni V et al., among south Indian population 91% prescriptions were found to have interactions [13].

Similar result could be seen the study done by Soherwardi S et al., [1]. In that study number of drugs per prescription ranged from 5 to 18, with an average of 7.8. The number of drugs per prescription was ranging from 3 to 10 with an average 6.00 in the study by Ahmad A et al., [15].

Out of 350 patients included in the study there were 156 males and 194 females. There was no significant difference between the frequency of DDI between males and females. In the study done by Salwe KJ et al., 62% were males and 38% were females but in the study by Ahmad A et al., among inpatients of a tertiary care hospital 53% patients were males and 47% patients were females [15,16].

For the evaluation of age wise distribution of DDIs, the participants were divided into five age groups. The age group with maximum number of patients and most number for interactions was found to be 61-80 years. There is a significant difference between average number of DDI among patients ≤60 years and patients >60 years. Soherwardi S et al., also have done an analysis of age wise distribution of DDIs with contradictory results [1]. In that study maximum number of patients were in the age group 56-65 years but there was no significant difference between the different age groups. In the study by Ahmad A et al., 71.7% patients belong to age group <60 years [15]. But Mallet L et al., had clearly demonstrated that elderly patients (>65 years) are most vulnerable for DDIs [17].

Maximum number of prescriptions had 6-10 drugs in list. Percentage of prescriptions with DDI and average number of DDIs was found to be increasing with increase in number of drugs. Average number of interactions were maximum (5.01) in the group >10. Similar pattern could be seen in the study done by Kulkarni V et al., [13]. In that study 85% prescriptions were in the group 6-10 and average number of DDI were maximum (9.40) in the group >10. Ahmad A et al., had categorised prescriptions in a different way (2-4, 5-7, and 8-10). Most number of prescriptions were in the group 5-7 [15].

Kulkarni V et al., had described three types of mechanisms underlying DDIs [13]. Apart from pharmacokinetic and pharmacodynamic mechanisms an unknown mechanism has also been described. A total of 42% interactions were pharmacokinetic, 24% interactions were due to pharmacodynamic mechanisms and 34% were considered as due to unknown mechanisms.

In the current study, 2.26%, 83.16%, 14.58% interactions were included in major, moderate and mild category respectively. It clearly shows that most number of interactions were of moderate severity. There is a significant difference in the frequency of major, moderate and mild interactions between pharmacokinetic and pharmacodynamic groups. Major interactions were more due to pharmacodynamic mechanisms and mild interactions were more due to pharmacokinetic mechanisms. Similar results could be seen in the study by Kulkarni V et al., with 2%, 70%, 28% DDIs in major, moderate and mild categories, respectively [13]. Ahmad A et al., demonstrated a different pattern with 31.65%, 53.95% and 4.38% DDIs in major, moderate and mild categories [15].

In the current study, drug groups most commonly involved in interactions were found to be antiplatelets, oral hypoglycaemic agents, bronchodilators, antibiotics, diuretics, insulin, statins, beta blockers, PPIs and NSAIDs. More or less similar observations could be seen in most of the studies referred [4,15]. Ahmad A et al., observed that NSAIDs, Antibiotics, PPIs and Corticosteroids were the most prone drug groups for interactions [15]. In the prospective study done by Bollu M the most common drugs involved were furosemide, ceftriaxone, paracetamol, atorvastatin, pantoprazole, atenolol and ampicillin [4].

Diseases most commonly associated with interactions were hypertension, diabetes mellitus, dyslipidaemia, coronary artery disease, cerebrovascular accidents, respiratory tract infections and chronic kidney disease. Most reference studies follow organ system wise categorisation of comorbidities. In the study by Ahmad A et al., fever, respiratory disease, cardiovascular disease, GI diseases and diabetes mellitus were found to be the prevalent comorbidities associated with DDIs [15]. Comorbidities of cardiovascular system, respiratory system, neurological system, GI system and endocrine system were most prone for DDIs in the study by Kulkarni V et al., [13].

The most common interventions for minimising the impact of DDIs were changing the timing of drug administration, monitoring for symptoms/signs/lab values/drug levels or both. Contradictory findings could be seen in the study done by Kulkarni V et al., and Bergk V et al., [13,18]. In those studies the most common management strategy was dose adjustments.

Analysis of the relation between duration of hospital stay and number of DDI showed a significant positive correlation with Pearson correlation value 0.406. Number of interactions was found to be increasing with increase in duration of hospital stay. Salwe KJ et al., had done a similar analysis and observed a positive correlation between duration of hospital stay and number of DDI [15]. There was an increase in number of DDI by a factor of 0.296 with one day increase on duration of stay in that study.

This study helped to find out the profile of DDIs, common drug groups and diseases associated with DDIs and the correlation between duration of hospital stay and number of DDIs among inpatients of a tertiary care hospital in Kerala.

The current study was conducted for a short duration. The investigator has not done any interventions to optimise therapy by minimising DDIs. No evaluation was done to assess impact of DDIs in the clinical condition of the patient. Extend of patient harm caused by DDIs were not studied.

Investigator bias might have happened. Investigator who is well aware of analysis plan herself had collected data. Interventional studies are needed to evaluate the clinical significance of DDIs and whether management of DDIs can reduce drug-related morbidity or mortality.

This study threw light upon the pattern and profile of drug-drug interactions among inpatients of a tertiary care hospital in Kerala. At least one DDI could be seen in 74.5% of prescriptions. Elderly people (>60 years) were found to be more prone for DDIs. Chance of DDI was found to be increasing with increase in number of drugs. Most number of interactions was of moderate severity, followed by mild category of interactions. Great effort need to be taken to minimise impact of DDIs for patients receiving antiplatelets, oral hypoglycaemic agents, bronchodilators, antibiotics, diuretics, insulin, statins, beta blockers, PPIs and NSAIDs. Diseases most commonly associated with interactions were hypertension, diabetes mellitus, dyslipidaemia, coronary artery disease, cerebrovascular accidents, respiratory tract infections and chronic kidney disease.

The easiest intervention for minimising the impact of DDIs is changing the timing of drug administration and close monitoring of the patient. There is a significant positive correlation between the number of interactions and duration of hospital stay.

Evaluation of the impact of DDIs on clinical and economic parameters needs to be done in upcoming research works. The effectiveness of different interventions in minimising patient harm due to DDIs is also a potential area for future research.