Friday, 23 August 2013
  0 Replies
  3.8K Visits
This summary is being cross-posted from the IAPHL discussion forum with many thanks! The International Association of Public Health Logisticians has recently been engaged in an interesting discussion about demand forecasting. Here are a few highlights from the email exchange. This includes demand forecasting best practices as well as common challenges encountered when trying to predict how much of a given commodity might be needed in the future. The conversation was launched when one participant inquired of the group, “why quantification of drug requirements is inherently imprecise, depending as much on art as on science.” Almost a dozen practitioners responded. There was broad consensus that demand forecasting will never be fool proof: “You are highly unlikely to accurately predict what the needs are for anything over a period of time (e.g. groceries for your family, medicines for a hospital, gas for your car, widgets bought by retailers).” Data can tell us there was a x% increase in demand for a given commodity over a certain number of years. That is the science part. The art is being able to say what will happen in the future: will it continue to increase x% next year? Will the increase be less because the program is nearing saturation? Or will it decrease because patients substitute a new product? There are countless instances when mathematical formulae alone will be unable to predict what might happen in the future. A few important demand forecasting lessons learned: · The biggest challenge to arriving at an acceptable forecast is accurate input data. This includes high quality and timely information about supply to facilities, stock on hand, back orders, and goods in the pipeline from suppliers. If these data are available, demand forecasting over a reasonable time horizon can get you most of the way there. · Even if the forecast exists and is solid, you still need someone to act on the data. An automated system can tell you what you need but someone still has to run the tender. · Good demand forecasts are still imperfect. In public health in particular, buffer stocks are essential and lifesaving. · It is not always unexpected variations in demand that result in stock outs; especially in low resource settings suppliers will not always meet their obligations to deliver on time. · One of the most important things is to have a responsive system that adjusts to changes in demand over time and minimizes lead time. Buffer stock will help but if the system is too unresponsive, massive wastage can result from, e.g., expired stock and changes in prescribing patterns. · Setting minimum and maximum stock levels requirements is not always optimal: http://tinyurl.com/kkrda4a · Distinguishing between need- and demand-based forecasts is crucial, as these can vary greatly. Need may not vary much given disease characteristics but demand is widely influenced by other factors. · Shortening the time horizon of the forecast improves accuracy. As do frequent and regular updates. To conclude, the IAPHL discussions participants recommended: · “We need to make our supply chains more flexible and agile, in other words more responsive to changes in demand in real time, to improve availability and efficiency. Shortening lead times and making funding more flexible would enable us to cut forecasting time frames. So by all means we need to improve forecasts, but on its own improved forecasting is not going to get us what we want; we need to look at the entire supply chain: data visibility, financing, procurement, and so on together.” · “[To encourage widespread adoption of best practices] publicize and share the many, many pockets of successes that we achieve everywhere - name a country and someone in the IAPHL distribution list can name a superb lasting and beneficial solution.” · Make logistics and its corollary, demand forecasting expertise, a recognized professional field. All the evidence suggests that with the right personnel, systems can work well, even in low resource settings. Here is a link to a great example in Tanzania. · “Ensuring that forecast accuracy be calculated along reporting cycles as data come from the facilities and other actual data becomes available can be very helpful in conveying the message of low quality of data.” · “Using the standard treatment guideline to work the client load backwards for most consumed commodities or regimens as a check for reported client load may be helpful. [So too can] relying on actual client load that have received commodities rather than clients that visited facilities from M&E. · “I suggest a one-time forecast which may be yearly or bi-yearly, and should be updated with REGULAR quarterly supply planning so that we can accordingly save lives and money. Supply Planning is the BEST way we can address, evaluate and implement our forecast; document the changes made in each supply planning for the next exercise and keep on improving.” Hope this helpful. Let me know if you have any questions. Best, Kira Thorien Associate Program Officer on assignment at Bill & Melinda Gates Foundation
There are no replies made for this post yet.