At the Mayo Health System in LaCrosse, Wisconsin, there are >1000 infant total parenteral nutrition (TPN) orders placed per year. It is the most complicated order that the pharmacy fills, so a recent peer-review article in Pediatrics moved a group of us to action at our center to buy or develop a TPN calculator. We did this because no stand-alone commercial calculators were available to us, and expensive electronic medical records typically do not include TPN calculators for neonatal patients. The new software includes decision support, and the orders are consistently legible. The physician performs fewer calculations, and there are no mathematical errors. This article examines the broader significance of providers having to write their own TPN software.
At the Mayo Health System in LaCrosse, Wisconsin, there are >1000 infant total parenteral nutrition (TPN) orders placed per year. It is the most complicated order that the pharmacy fills, so a recent peer-review article1 in Pediatrics moved a group of us to action at our center to buy or develop a TPN calculator.
This article by Lehmann et al of Johns Hopkins, “Preventing Provider Errors: Online Total Parenteral Nutrition Calculator,” relates that the John Hopkins Medical Center was able to take paper ordering of neonatal TPN and make an electronic ordering program within 3 weeks by using in-house talent.1 More importantly, they showed what I suspected: value to patients. They were able to reduce errors from 10.8 per 100 TPN paper orders to 1.2 errors (89% reduction) by using this electronic format. Two thoughts immediately came to mind: if they can do it, why can't we? I did my neonatology fellowship at Cornell University, and we always wrote neonatal TPN orders electronically; thus, I wondered if I could just get the program from them. Nevertheless, I finished my fellowship at Cornell–New York Hospital Medical Center 16 years ago, and I remember from my last visit to them that they had not only purchased an expensive electronic medical record (EMR) system but had merged with Columbia University Presbyterian Medical Center, so I was not sure if friends at Cornell could just send me a neonatal TPN program. I began to wonder how the Cornell software would integrate into the expensive EMR system that they had purchased a few years ago. Maybe the new EMR system at Cornell included a TPN calculator. Maybe most EMRs do not include TPN calculators. I made some phone calls, and I learned from the director of a large academic NICU (he prefers not to be named) what occurs at his center: “NICU staff and residents still submit TPN orders on paper despite an expensive EMR, and lunchtime always means numerous phone calls to prevent errors because of math errors or handwriting that cannot be deciphered by a nurse or pharmacist.”
The second thought I had was that reimbursement rules largely don't recognize information technology–related capital expenditures, but if I assembled a team to write a neonatal TPN program for our patients, we could theoretically do it on salaried, already paid time; thus, we could get more value for the “health care” dollar of our patients. In fact, not only is medical software expensive, but we cannot always buy what medical software we need, nor can we fix problems or implement advances without spending more money on updates or new vendors. The John Hopkins work in Pediatrics relates how, if we write it ourselves, we can “find it, fix it, and forget it” and get good value for very little financial cost to our patients. This targeted surgical approach (cutting out the problem and replacing the defective process) that was applied by Lehmann et al, in conjunction with the significant domain knowledge of his team in the area of TPN, allowed the design of an application that did reduce errors. If we wrote such a program, we would have to follow Title 21 Code of the federal regulations, which requires software systems to ensure accuracy, reliability, and consistency. We thought that if we found a problem, we would fix it and be cautious that the fix would not generate an error risk before we “forgot it.”2
It became apparent that I could not buy a neonatal TPN program for our NICU. I called Alfred N. Krauss, MD, at Cornell and learned that a second-year pediatric resident wrote a homegrown TPN program by using Microsoft Excel (Seattle, WA). I was amazed that a pediatric resident had to write a calculator after Cornell had purchased an EMR system for millions of dollars (Alfred N. Krauss, MD, written communication, January 4, 2005). I was not even sure if the Mayo Health System would just let me begin using Cornell's neonatal TPN calculator, but Dr Krauss told me that he could not just give the program to me.
Thus, with the agreement of our medical director and our organizational spokesman for system reliability and patient safety (Stephen P. Shultz, MD), we followed the John Hopkins example and made our TPN calculator by using Microsoft Excel. We (myself, Jose Yuvienco, MD, and Karen Olson, RN [of our neonatology department], Scott W. Mihalovic, PharmD, RPh, and John K. Johnson, RPh [of our pharmacy], and Gina L. Howlett, BSN, RN, and Jerrilyn R. Hendrickson, AB [of our information-services department]) did this because no stand-alone commercial calculators were available to us, and expensive EMR systems typically do not include TPN calculators (especially for infants). I found this hard to understand given the results of a Harvard study published in 2001, which found that 79% of potential adverse drug events (ADEs) are preventable, and the potential ADE rate is highest among neonatology inpatients when compared with an older population.3 The US Food and Drug Administration, World Health Organization, and International Committee on Harmonisation define an adverse drug reaction as a “response to a drug which is noxious and unintended and occurs at doses used in man for prophylaxis, diagnosis, therapy or modification of physiologic functions.”4 In fact, although some studies (Boston, MA)5 can show only a 17% decrease of serious preventable ADEs, other studies have shown an 81% decrease of serious preventable ADEs because of a computerized provider order entry (CPOE) system.6 In fact, I wonder if that is not why Lehmann et al wrote their own TPN-calculator program?
We used ranges based on recommendations for TPN published by the American Academy of Pediatrics (AAP), nutritional needs of the preterm infant, and some recent nutritional updates for premature infants.7–11 I could not help wondering if a Web-based calculator from the AAP would not be a good addition to their textbook? After all, I thought, why do TPN orders have to be so dependent on which medical center a given infant is born? Would not how much protein, multivitamins, and trace elements in the TPN be based largely on the birth weight and gestation? The software could give decision support (eg, how much cysteine for a given number of grams of protein) or an alert to decrease trace elements if the infant has liver disease by using the same sound medical evidence and consensus already contained in the Pediatric Nutrition Handbook.7 The AAP experts would have the best understanding of how much decision support the average doctor might need to safely and efficiently compose TPN for most neonatal patients.
Do I think there will be demand for such a calculator? Yes. In a recent survey describing neonatal nutrition-design techniques for very low birth weight infants at 8 neonatal centers in North Carolina, the authors document the low frequency with which NICUs use electronic ordering systems to design parenteral nutrition.12 It has been estimated that 750000 neonatal TPN orders are prescribed each year in the United States. How and what percent of neonatal units use an electronic TPN calculator could be ascertained by an AAP or even a Vermont-Oxford survey, but why should the state of North Carolina be atypical? I wrote to the editor of the AAP's Pediatric Nutrition Handbook, who told me that the expectation is that most doctors are making these programs (Ronald E. Kleinman, MD, written communication, June 22, 2005).
The new TPN software that we created is working wonderfully, and multiple peer-review articles in the medical literature discuss the importance of both quantitative and qualitative assessments of new ordering systems. Lehmann et al also performed a quantitative study, so I will give the qualitative perspective of an electronic TPN calculator (Fig 1). The orders are legible and have reduced “the risk for error.” When my partner and I fill out TPN orders for an infant, we enter data into 24 fields, and the computer calculates the other 35 field values, with no mathematical errors. This saves us laborious, tedious calculations that we ordinarily would have to do for several infants per day. On any given day, not only can we write a TPN order faster and with more accuracy and safety, but we can also review it and adjust components if necessary (eg, a few milliliters more of fluid or a little more dextrose) without starting all over again. The computer helps us to modify the TPN solution as enteral feeds increase and calculates TPN changes as we adjust arterial-line fluid rates. The computer saves the order; the next day, any needed changes are made in the saved copy, so the order takes even less time. While creating our homegrown software, we made sure that there were no fragmented displays of the orders (one screen shows all the TPN components).13 The new software delivered exactly what I would expect from a well-designed CPOE system, which provides support to us in 2 principal ways:
It standardizes repetitious (high-volume) tasks: errors in information transfer processes can be reduced through the use of guided and automated data entry and validation and standardized terminologies to provide legible and verified orders and communications. Commercial CPOE systems use this strategy (if nothing else, all orders will be legible).
It simplifies complex (error-prone) tasks: errors caused by cognitive overload can be reduced through calculation, reminders, and rule-based decision support at the bedside. Decision support provides data to the provider to optimize the selection and specification of medication for ordering.
We presented our new program at the Mayo Clinic Formulary and Pharmacy Retreat as an example of patient benefit at low cost and offered to let other Mayo Clinic practice sites use or create their own software from already existing paper orders. The program was well received by the group, and they particularly liked the edits and reminders that we built-in for patient safety. We enter the TPN orders into a computer, and subsequent generation of alerts and reminders means that the pharmacist does need to call us. The nurses love the legibility of the computer TPN orders. We not only observed for any errors that we want our software to prevent, but we looked and surveyed staff for any errors that might be caused by our new software; we did not find any errors caused our new TPN software. Alarms were not ignored, there were no computer crashes (and we still had the ability to write the same orders on paper), and no orders were entered into the wrong medical record. In our NICU, one doctor per day orders all the TPN. We have a dedicated computer for the 2 neonatologists, so there is always a clear log-on/log-off status for each physician. Our development time was a few weeks, and we did this without sacrificing our other patient, academic, and administrative duties.
After our new TPN calculator was made and being used, I wrote to Lehmann, and he sent his TPN program to me free of charge within an hour of my e-mail (Christoph U. Lehmann, MD, written communication, July 20, 2005). Why didn't I just do that in the beginning?
Is there a broader significance to all this? We think there is. On February 19th, 2005, The New York Times reported that our government would like the health care industry to move toward electronic ordering of medications and an EMR.14 The thinking is that as 40% of the health industry becomes electronic, the societal dividend will be a 10% reduction in medical errors. In fact, “if steps are not taken soon to make it happen, the government will probably impose a solution.”14
In the June 2005 issue of Hospitals & Health Networks, the Journal of the American Hospital Association discusses the thoughts of David Brailer, MD, the federal government's national health information technology coordinator, at the Healthcare Executive Forum: “Brailer was adamant that government should not decide the technical infrastructure that will allow interoperability, or the ability for patient information to seamlessly flow from one provider or institution to another within 10 years. Brailer suggests the government should encourage the industry to do it, and ensure that the ultimate design is fair. ‘This has got to live in the private sector, but I think regulation is always an option… that might be where we end up.’”15
I do not think that a government mandate is necessary, and I think discussion on the Maine lobster industry is a good example for pediatrics and medicine. In l976, the US Congress passed legislation called the Fishery's Conservation and Management Act. The federal government began to have control over the fishery with a well-intended law to protect US waters from foreign fishing fleets. It also empowered the National Marine Fishery Service to analyze the lobster fishery in the Gulf of Maine. The federal government thought that too many lobsters were being caught before making eggs and thus predicted that federal regulators would need to play a very active role or the lobster population would dwindle and the industry would collapse. Over the years, however, the lobster populations in the Gulf of Maine actually grew. The federal government had not noted a conversation practice that Maine lobstermen had long prided themselves as being homegrown: if they thought a lobster was a breeder, they put it back into the water. The government regulators missed this conservation practice.16
Is there any chance that physicians and the health care industry are proving to be like Maine lobster fisherman? I believe so, and with the latest estimate for the cost of electronically connecting America's health care system running approximately $300 billion US dollars, the return on investment will depend on physicians to actively assess and create safer CPOE products.17 Consider Wisconsin's Peridata as another good example. Peridata is the result of a unique public-private partnership in Wisconsin called PC-log.18 Peridata is affordable (approximately $5000 to start up) and will allow almost all of the Wisconsin hospitals with midwifery and obstetrics to compare and benchmark patient outcomes within the state of Wisconsin. Peridata allows electronic reporting of birth certificates. Peridata creators are working on electronic reporting to Vermont-Oxford, and I am recommending to my Wisconsin colleagues that Peridata follow the example set by Lehmann et al and create a reliable, Web-based neonatal TPN calculator. Thus, I believe a federal mandate for health care to be computerized would be premature.
In summary, Vermont-Oxford, Peridata, and the excellent work by Lehmann et al and those at Cornell University are good examples of why I think it is important that pediatricians be actively involved in the creation of new software and the purchase of existing CPOE systems. Otherwise, we may wake up to find a federal regulator in the neonatal TPN.
- Accepted July 29, 2005.
- Address correspondence to Dennis T. Costakos, MD, FAAP, Department of Pediatrics (Neonatology), Franciscan Skemp-Mayo Health System, 700 West Ave S, LaCrosse WI 54601. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Lehmann CU, Conner KG, Cox JM. Preventing provider errors: online total parenteral nutrition calculator. Pediatrics.2004;113 :748– 753
- ↵Berger RG, Kichak JP. Computerized physician order entry: helpful or harmful? J Am Med Inform Assoc.2004;11 :100– 103
- ↵American Academy of Pediatrics. Nutritional needs of the preterm infant. In: Kleinman RE, ed. Pediatric Nutrition Handbook. 5th ed. Elk Grove, IL: American Academy of Pediatrics;2004:36
- Costakos DT, Porte M. Did controversies concerning vitamin k and the newborn cover all the controversies [letter]? Pediatrics.2004;113 :1466– 1467
- ↵Brion LP, Bell EF, Raghuveer TS. Variability in the dose of intravenous vitamin e given to very low birth weight infants. J Perinatol.2004;25 :139– 142
- ↵Ash JS, Berg M, Coiera E. Some unintended consequences of information technology in health care: the nature of patient care information system-related errors. J Am Med Inform Assoc.2004;11 :104– 112
- ↵Lohr S. Health industry under pressure to computerize. New York Times. February 19,2005:C1
- ↵Gearon C. Everybody supports the EHR. Now what? Hosp Health Netw.2005;79(6 suppl) :4– 6
- ↵Corson T. Waiter! There's a federal scientist in my lobster bisque. The Wall Street Journal. May 12,2005:A17
- ↵Gearon C. Show me the seed money. Hosp Health Netw.2005;79(6 suppl) :6 , 8
- Copyright © 2006 by the American Academy of Pediatrics