Economic Impact and Cost Benet of the Inpatient Clinical Pharmacist Interventions in Cancer Care Units

1 Background: Clinical pharmacists play a key role in ensuring the optimum use of cancer medicines. Yet, 2 the economic benefit of this role has never been assessed in Qatar. 3 Aim: To evaluate the overall economic impact of clinical pharmacist interventions in the main cancer care 4 setting in Qatar. 5 Methods: From the public healthcare perspective, this was an analysis of the total economic benefit and 6 a cost-benefit analysis of the clinical pharmacy interventions. As a study sample size, patient records in 7 March 2018, July/August 2018, and January 2019 were retrospectively reviewed at the National Center 8 for Cancer Care & Research (NCCCR), Qatar. The total benefit from interventions was the total of the cost 9 avoidance due to preventable adverse drug events (ADEs) plus the cost savings associated with 10 therapeutic interventions. The interventions cost was based on salary and increased cost due to 11 therapeutic interventions. The cost-benefit analysis results were presented via net benefit and benefit‐ 12 to‐cost ratio measures. 13 Results : Total of 1,352 interventions occurred during the 3-month follow-up period. The total benefit was 14 QAR 196,010,360 (USD53,834,206), constituting cost avoidance of QAR 194,764,534 (USD 53,492,040) 15 and cost savings of QAR 1,245,826 (USD 342,166), mostly due to recommending additional medications 16 and the medication dose reduction. The benefit-to-cost ratio was 174:1 and the annual net benefit was 17 QAR 779,539,440 (USD 214,100,351). Sensitivity analyses confirmed the robustness of results. 18 Conclusion: The clinical pharmacist intervention is a cost-beneficial practice in the NCCCR setting, 19 associated with ADEs prevention and substantial economic benefits, including relative to the interventions 20 cost.

The cost of the intervention was calculated based on the sum of the salary of clinical pharmacists, added 111 to any increased cost of therapy associated with the administration of intervention. The basic monthly 112 salary of a clinical pharmacy in Qatar is estimated to be around QAR20,000 (USD5,493). For the current 113 study, however, calculations were based on an overestimated monthly salary of QAR30,000 (USD8,239). 114 Details of the cost of interventions calculations can be seen in Supplementary Information S1. As per the 115 duration of study follow up, the cost of intervention over 3 months was calculated. 116

Cost savings 117
Cost savings based on the clinical pharmacist interventions were the reduced cost of therapy associated 118 with therapy changes due to the intervention. Details of the cost savings calculations can be seen in 119 Supplementary Information S1. As per the duration of study follow up, the cost saving over 3 months was 120 calculated. 121

Cost avoidance 122
Cost avoidance was the cost avoided by eliminating the occurrence of ADEs as a consequence of the 123 clinical pharmacist interventions [15]. Based on the method of Nesbit et al., the likelihood of an ADE in 124 the absence of the intervention was determined at 0 (none), 0.01 (very low), 0.1 (low), 0.4 (medium), or 125 0.6 (high). The description of Nesbit et al. method is shown in Supplementary Information S2. The cost of 126 an ADE was calculated based on the conservative assumption that an ADE will lead to an additional 2 days 127 of hospital stay in the relevant unit, which is consistent with previous studies [16]. Further details of the 128 cost avoidance calculations can be seen in Supplementary Information S1. As per the duration of study 129 follow up, the cost avoidance over 3 months was calculated. 130 Cost-benefit analysis 131 8 Results of the cost-benefit analysis were presented in terms of the total benefit, the benefit-to-cost ratio, 132 and the net benefit of the intervention, which was calculated in 3-monthly and annual values. Cost is the 133 cost of the interventions, while the total monetary value of the benefit of the interventions was calculated 134 as the sum of the cost savings and the cost avoidance associated with the interventions. 135 The study was conducted from the perspective of the public NCCCR hospital. Thus, only direct medical 143 costs were considered in the analysis. 144

Cost inputs 145
The monetary values of resources were calculated, where the cost of medications, non-medication-based 146 resources such as laboratory and diagnostic tests, and hospital stay were obtained via the pharmacy and Square tests were used to detect any significant differences among the three sample size groups, i.e. 171 March 2018, July/August 2018, and January 2019. 172

Sensitivity analysis 173
One-way and multivariate sensitivity analyses were used to assess the impact of uncertainty in the base-174 case values of variables in relation to the costs and ADEs probabilities, to enhance the 175 robustness/generalizability of the study conclusion. 176 One-way sensitivity analysis was performed to target values of individual uncertain input variables to 177 analyze the effect of uncertainty on the study conclusion. Targeted uncertain inputs included the clinical 178 pharmacist salary and the cost of the ADE, using an assigned ±20% uncertainty range. 179 Multivariate uncertainty analysis was performed by targeting several underlying uncertain probabilistic 180 inputs in the economic model before re-running the model several times, for a distribution of the model 181 results to be generated. Here, the probabilistic inputs of interest were the probabilities of the avoided 182 ADEs set by the expert panel, using an assigned ±15% uncertainty range for any probability. There were no significant differences between the study groups except with regards to the ward type 197 (Table 1). 198 Supplementary Information S3 presents a summary of the categories of the interventions types, including 199 study examples, and the associated average probability of avoided ADEs as per category. 200

Cost of intervention 202
Based on interventions performed in the current study, a clinical pharmacist performed an average of 250 203 interventions over the 3-month study period. Based on total number of interventions in the current study, 204 6 clinical pharmacists were needed, where the overall salary of the clinical pharmacists was calculated to 205 be QAR180,000 (USD49,437). For the added cost associated with therapeutic interventions for the DRPs 206 over a 3-month period, this was QAR945,500 (USD259,681). The total cost of the intervention over the 3-207 month study period is, therefore, adding to QAR1,125,500 (USD309,118). The added cost with 208 interventions as per different intervention type categories can be seen in Table 2. 209

Cost saving 210
The overall cost saving was QAR1,245,826 (USD 342,166). Cost saving as per different intervention type 211 categories can be seen in Table 2. is the estimated probability with each of 9 interventions. The overall cost avoidance due to the 12 interventions over a 3-month period was QAR194,764,534 (USD53,492,040). Table 2 summarizes cost  218 avoidance associated with each type of intervention category. 219 Cost-benefit analysis 220 The total benefit due to the interventions over a 3-month period was QAR196,010,360 (USD 53,834,206). 221 The benefit-to-cost ratio  Table 3. 225

Sensitivity analysis 226
The outcome of the economic model was insensitive to uncertainty in the clinical pharmacist salary and 227 in the cost of the ADE. 228 Multivariate sensitivity analysis demonstrated that there is a 100% probability that the pharmacist 229 intervention is associated with positive a 3-monthly net benefit and total benefit (Fig.1a) and an annual 230 net benefit (Fig. 1b). Furthermore, as illustrated in Fig. 1c, the analysis showed that there is a 100% 231 probability that the pharmacist intervention is associated with an >1 benefit to cost ratio. 232 Input uncertainties and sampling distributions used in the one-way and multivariate sensitivity analyses, 233 and the details of their outcomes, are shown in Table 4. 234 A regression Tornado analysis revealed that the main driver of the outcome was the cost of ADE, followed 235 by the 0.3 probability of avoided ADE (Fig. 2). Clinical pharmacists also play an important role in disease-related problems. Indeed, the deployment of 289 clinical pharmacists in clinical rounds led in our study to recommend interventions related to laboratory 290 testing such as therapeutic drug monitoring, diagnostic, and culture tests, which resulted in a cost saving 291 of QAR23,807 (USD6,539) and a cost avoidance of QAR45,025,307 (USD12, 366,192). 292 This study is not without limitations. The cost-benefit might be over or underestimated due to the nature 293 of the retrospective study design, which has the inherent limitation of missing data, especially as daily 294 reporting of interventions is not mandatory. Furthermore, the clinical outcome of patients, wherein 295 clinical pharmacists intervened, were only analyzed using the probabilities of avoidable ADEs, which may 296 not reflect the real impact of interventions beyond the ADEs. In addition, the cost of ADEs was calculated 297 with the assumption that it is equal among the different ADEs, universally associated with an additional 298 two days of hospital stay. As already discussed in methods, however, this is justified, especially since it is 299 impossible to follow up patients until the end of intervention consequences that never took place. Also, 300 in our study, a 3-month follow-up duration may not accurately represent the real economic benefits over 301 one year. However, given how the 3-month follow up was selected as a sample size, as already discussed 302 above, and given the lack of significant differences in patient characteristics between the 3 months of the 303 follow-up duration, there is no reason to assume that the remaining months of the year will differ and 304 that results from the study's 3-month follow-up cannot be extrapolated to calculate an annual benefit. 305 Finally, while the comprehensive one-way and multivariate sensitivity analyses performed would 306 considerably increase the robustness and generalizability of results beyond the study setting, our results 307 will need to be considered with caution by other settings, given that the resource utilization and costs can 308 considerably differ.