Advance patient safety with the right prescription. First time.
Our health economics analysis evaluates the potential cost savings and time efficiency gains associated with Touchdose, and highlights the methodologies used to calculate the figures for secondary care.
We have used baseline error rates for both paediatric and adult prescribing derived from large, multisite studies, and incorporated findings from a recent Imperial College London study, published by BMJ Quality & Safety (1), that shows an 84% reduction in the odds of a prescribing error with Touchdose.
In addition, the analysis quantifies non-cash releasing savings from reduced prescribing time with Touchdose. Drawing on findings from a user-testing study, and recognising that costs vary by prescriber seniority, it uses specific NHS cost data for different prescriber roles, adjusted for location-based cost variations.
Read the evaluation below.
__________________________________________
If you'd like more information on this health economic evaluation, or to discuss Dosium further, please reach out to our team: hello@dosium.com
__________________________________________
__________________________________________
Economic evaluations, such as cost-consequence analyses, provide a structured framework to quantify the health, cost, and resource implications of implementing digital health technologies such as Touchdose. By presenting disaggregated costs and outcomes, these analyses support decision-makers in making informed choices within their specific healthcare context.
In this health economic analysis, the potential cost savings and efficiency gains from Touchdose are explored by examining its ability to reduce prescribing errors and prescribing time in inpatient settings. The analysis evaluates both inpatient and discharge medication orders, using baseline error rates from the literature, prescriber grade breakdown, and NHS cost estimates to provide a detailed assessment of Touchdose’s economic impact. This approach ensures alignment with NICE guidelines and offers robust evidence to support the system's value in improving safety and efficiency.
__________________________________________
To calculate the cost savings associated with a reduction of prescribing errors by using Touchdose, the following information was required: number of medication orders/prescriptions written, baseline error rates, burden of harm from errors that lead to an increase in length of stay, cost of additional bed days, and the estimated effect size of implementing Touchdose.
How each of these figures were derived will now be detailed in turn.
The number of medication orders are based on actual or estimated numbers based on national / published data to calculate the potential cost saving.
National figures are derived from the total number of secondary care admissions in England, as reported annually by NHS England (2). This figure was then multiplied by 4.78, the mean number of medication orders per admission, based on findings from a study of 20 hospitals, evaluating 26,019 patients (3).
If calculating cost savings at a Trusts or department level, the numbers provided by a Trust can be used to provide a more localised evaluation. Ideally this will have been provided as two figures, one for paediatric patients and one for adult patients.
The calculations can therefore be conducted for either paediatrics only, paediatrics and adults (then combined), or mixed patient populations (where the number of orders by group has not been provided).
A baseline error rate for paediatric prescribing of 13.2% has been used as reported in a study of eleven wards from five hospitals, evaluating 2,955 medication orders (4). The baseline error rate for adult prescribing is 8.8%, taken from a study of 20 hospitals, evaluating 124,260 medication orders (3). These studies were selected for their large sample sizes, rigorous methods, and inclusion of observational data rather than incident report data. They also utilised multiple hospital sites for data collection, further enhancing the reliability and generalisability of the reported rates.
Where the numbers of prescriptions/medication orders for adults and paediatrics has not been defined by a trust the baseline error rate of 8.8% has been used.
The estimated annual number of errors will be calculated by multiplying the number of prescriptions or medication orders by the baseline error rate percentage.
Where a baseline error rate or annual number of errors per annum is available for a specific department or Trust, such as from audits, service evaluations, or research, this can be used as an alternative to the rates reported in the literature (3,4).
However, it should be noted that baseline error rates derived from local incident report data for a department or Trust can be significantly understated. This is highlighted by a recent study that identified 11,302 clinical prescribing errors (5). The study found that staff detected only 26% of potentially serious prescribing errors and reported just 1.1% of these to the incident reporting system. Furthermore, 82% of errors associated with actual patient harm went undetected, and 90% had no corresponding incident report (5). Therefore, baseline data should be based on more robust data collection methods. In the absence of such local data, rates reported in the literature should be used.
The percentage of prescribing errors likely to cause moderate or severe harm is reported as 51.6% and 7.25% respectively (3). Whilst it could be assumed that prescribing errors classified as likely to cause moderate or severe harm would also result in adverse drug events or reactions, it is important to acknowledge that "likely to cause harm" does not necessarily mean that harm will actually occur. Therefore, to provide a more conservative estimate for this economic evaluation, only the percentage of prescribing errors likely to cause severe harm (7.25%) will be used to calculate the burden of harm from errors that lead to an increase in length of stay.
A study of 3,695 secondary care patient episodes identified that 545 had one or more adverse drug reaction, of these 26.8% had an increased length of stay, with a median of 4 additional bed days (6). This is the equivalent of 1.06 additional bed days for each adverse drug event.
The unit cost per bed day was £406, as per the NHS National Cost Collection 2021/2022 (7), which provides the figure for a regular day or night admission.
The Touchdose user testing study evaluated 24 clinicians and 240 medication orders comparing usual practice with use of Touchdose. There was a statistically significant reduction in the odds of a prescribing error by 84%, from 28.3% to 6.6% (odds ratio 0.16, confidence interval 0.06 - 0.43, p<0.01) when using Touchdose compared to usual practice (1).
The prescribing tasks presented to participants in the study were moderately complex, however, the tasks included reasonable dosing requirements or adjustments that are reflective of every day prescribing tasks, such as the need to apply a maximum dose, ideal body weight (for obese patients for relevant medications) or to calculate the patient's body surface area.
Using an example trust that has 59,317 paediatric and 1,682,135 adult medication orders per year the following potential cost savings have been calculated (8).
The introduction of Touchdose has the potential to provide an annual cost saving of £216,225 for paediatric orders and £3,722,872 for adult orders by reducing bed days associated with prescribing errors. This is a combined total of £3,939,097. See Table 1 for a detailed breakdown of how the cost savings were calculated.
.png)
Using national figures for England, where the estimated number of medication orders per year is 83,937,412, the following potential cost savings have been calculated:
The introduction of Touchdose has the potential to provide an annual cost saving of £178,302,941 by reducing bed days associated with prescribing errors.
Before the introduction of Touchdose, the baseline prescribing error rate was 8.8%, resulting in 7,386,492 errors per year, of which 535,521 errors per year were likely to lead to severe harm. These severe harm errors were associated with an estimated 574,078 additional bed days, translating to a cost of £233,075,740 in additional bed days.
With Touchdose, the prescribing error rate is projected to reduce to 1.41%, leaving a residual 1,735,826 errors per year, of which 125,847 errors per year would likely lead to severe harm. This reduction is expected to decrease the associated additional bed days to 134,908, reducing the cost of additional bed days to £54,772,799.
This results in a total annual cost saving of £178,302,941.
__________________________________________
To enable calculation of the time saving and non-cash releasing savings from a reduction in prescribing time using Touchdose the following information was required; number of medication orders/prescriptions issued, the time saving in seconds associated with the use of Touchdose, prescriber split proportions, costs associated with prescriber time.
How each of these figures were derived will now be detailed in turn.
The number of medication orders are based on actual or estimated numbers based on national / published data to calculate the potential cost saving.
National figures are derived from the total number of secondary care admissions in England, as reported annually by NHS England (2). This figure was then multiplied by 4.78, the mean number of medication orders per admission, based on findings from a study of 20 hospitals, evaluating 26,019 patients (3).
If calculating cost savings at a Trusts or department level, the numbers provided by a Trust can be used to provide a more localised evaluation.
The time saving estimate has been taken from the Touchdose user-testing study that evaluated 22 users and 220 medication orders* comparing usual practice with the use of Touchdose (1). There was a median time saving of 35 seconds per medication order when using Touchdose. For more complex prescribing decisions where the time required to determine a dose is extended, the time saving (for the upper quartile) was 56 seconds per medication order when using Touchdose.
For the purpose of this economic evaluation, to provide a more conservative estimate, the median time saving of 35 seconds will be used. This figure has been used to calculate the number of seconds, hours or days saved based on the number of prescriptions/medication orders annually (the number of orders may be at a ward, department or trust level).
*The sample size for ‘time to prescribe’ was reduced to 22 users and 220 medication orders due to recording failure during 2 participant sessions.
As the seniority of the prescriber will influence the cost of the time to the organisation, the grade/band of the prescriber has been taken into consideration. This is particularly important as in most hospital settings the majority of prescribing is done by junior doctors, in particular those in foundation years 1 and 2.
The breakdown of prescriber grade as percentages have been taken from a study evaluating 20 hospitals and 124,260 medication orders (3). The percentages have been derived by taking the number of orders per prescriber group and dividing this by the total number of orders. The prescriber split percentages are available in the appendices, Table A2.
Where trusts can provide their own data detailing prescriber split, this can be used as an alternative to the prescriber split derived from Ashcroft et al (3).
Prescriber (staff unit) costs were obtained from the NHS Benefits Calculator Tool using 2023/2024 data (9). For medical prescribers, the average cost for each grade was taken from this tool. For non-medical prescribers (e.g. nurses and pharmacists), an average cost across bands 6-9 was used. This was necessary because the reporting on prescriber splits did not specify the band or seniority level for non-medical prescribers. As a result, an average cost for bands 6-9 (calculated using the average for each band) was applied. Band 5 costs were excluded, as nurse prescribers are typically at least senior staff nurses (band 6 or higher), and pharmacists qualify at band 6. If estimating costs for trusts that provide high cost area supplements, the appropriate percentage uplift will be included as either 5% (Fridge), 15% (Outer London) or 20% (Inner London).
Mapping the prescriber groups used in the Ashcroft et al. (3) study to NHS cost groups was required to estimate the financial impact of different prescriber roles. While Ashcroft’s clinical categories do not perfectly align with NHS cost structures, this approach provides a pragmatic and reasonable way to assess the costs associated with prescribing practices. Despite the limitations in directly matching these categories, this mapping still allowed for a more accurate estimation of cost compared with using an average cost across all prescriber groups. The mapping of the prescriber groups is available in the appendices, Table A3.
Using an example trust that has a total of 1,741,452 medication orders per year, the potential cost savings have been calculated based on this figure (8). The example trust is based in Central London and therefore a 20% high cost area supplement has been applied to the costs.
The total number of hours saved annually across all prescriber groups is 16,930.78, equivalent to 705.45 days saved on prescribing. This reduction in prescribing time translates to a total cost saving of £463,425.61 per year, based on the calculated time savings for each prescriber group. See Table 3 for a detailed breakdown of the time and cost savings per prescriber group.
Using national-level data, the potential cost savings from a reduction in prescribing time have been calculated based on a total of 83,937,412 medication orders per year.
The total time saved across all prescriber groups is 1,398,957 hours annually, equivalent to 58,290 days, which translates into a total cost saving of £22,336,962 per year. See Table 4 for a detailed breakdown of time and cost savings by prescribed grade.
Health economic analyses are generally based on assumptions and extrapolation, which may not fully capture the complexities of real-world prescribing and cost scenarios.
A limitation of the literature used to determine baseline error rates is that these studies included paper-based prescribing. As there is limited generalisable baseline prescribing error literature available since the widespread adoption of electronic prescribing, these studies were used despite their limitations. The few studies that do exist are small scale, including two English studies that reported no statistically significant difference in error rates since the introduction of electronic prescribing (10,11).
In inpatient settings, prescribing medications is often a collaborative process. Senior medical staff, such as consultants, typically make the decisions and delegate the actual prescribing to junior staff. However, prescribing data from electronic records only identifies the individual who created and signed the prescription. As a result, the time and cost savings highlighted in this analysis are likely underestimated, as they don't fully account for the involvement of multiple team members.
A further limitation of this analysis is that the prescriber split information is based on data from adult and/or mixed patient populations, which may not accurately reflect prescribing practices in paediatrics. In paediatric settings, prescribers are often more senior, meaning their associated costs are higher. This discrepancy may lead to an underestimation of the true cost impact in paediatric care.
__________________________________________
__________________________________________
If you'd like more information on this health economic analysis, or to discuss Dosium further, please reach out to our team: hello@dosium.com
