Use of a codified medication process for documentation of home medications

Abstract

To determine the quality and completeness of the list of home medications documented by nurses using a codified process, authors conducted a comparative study of home medications using a non-codified and codified process for documentation of required data fields including drug, dose, route of administration, frequency, and schedule. Each documented home medication (DHM) was evaluated based on the ability to convert to an inpatient medication order. The home medication was classified as non-convertible if one or more of the required data fields were missing, inaccurate, or incomplete. The study compared 176 patients with 1618 DHM in the non-codified group to 94 patients with 646 DHM in the codified group. All DHM could be converted to inpatient orders for 70% of the patients in the codified group compared with 42% in the non-codified group. Based on each DHM, the codified process resulted in 92% of the DHM being able to convert to inpatient orders compared with 82% for the non-codified process. Authors conclude that use of a codified process to document home medications has the potential to increase the number of complete drug entries and in the number of patients with a DHM list in which all of the medication entries have all of the dosing information.

The Joint Commission National Patient Safety Goals include as an objective to accurately and completely reconcile medications across the continuum of care.¹ Reconciliation is the process of comparing the medications that the patient has been taking at home with new sets of medications that the patient receives on admission to the hospital and to those at the time of discharge. The purpose of the reconciliation is to avoid errors of transcription, omission, duplication of therapy, drug–drug and drug–disease interactions.

To accomplish this goal, on admission a complete list of medications that the patient was taking at home must be created. Each documented home medication (DHM) should include the drug, dose, route of administration, and frequency.²

The method in which the documentation takes place varies by institution.³ Options include using an electronic form with a structured data entry or codified process,^4–6 handwritten forms with a free-text or non-codified process,^{4 7–16} or a combination of both.^{17 18}

Most electronic medication lists use a structured data entry or codified process, meaning that the data elements of the medication order (ie, drug, dose, route of administration, and frequency) can be recognized and used by other electronic systems. Depending upon the level of functionality of the system being used, DHM can be gathered from other electronic sources such as outpatient medication records or discharge medication lists.^{5 6 13}

Another option for creating a codified DHM list is to require the healthcare provider to enter all of the data elements into electronic form using a codified process wherein standardized multiple choice menus are used for each of the data elements. With this system drugs are chosen from a national drug databank list which includes both legend and non-legend (over-the-counter) medications. Once the drug is selected, information for each of the other data fields (dose, route of administration, and frequency) are shown using a knowledge-driven medication order functionality with the most common, common, least common, and other choices highlighted using different icons.

Medication lists created using a non-codified process may be a source of errors when reconciling admission medication orders. In general, the data elements for each DHM are written free-form by the healthcare provider. This means that there is no specific format for the data or control over the validity of the data. Studies comparing DHM and admission medication orders have reported discrepancies related to incorrect drug, dose, and frequency ranging from 22% to 62.3%^{4 7 11 12 14} of all order discrepancies. Whether the discrepancies were the result of incomplete or inaccurate patient recall of their medications, inaccurate transcription of the information by the healthcare provider, or changes in therapy upon admission has not been evaluated.

Studies at one hospital have shown that the incidence of admission medication reconciliation discrepancies is less using a codified DHM list.^{13 17} As part of a larger process improvement project and upgrade of the existing electronic medication record system, Advocate Good Samaritan Hospital, Downers Grove, Illinois changed from a non-codified process to a codified process for the documentation of home medication. The codified DHM list was fully compatible and integrated with the inpatient computerized provider order entry system thus allowing for the conversion of DHM orders to inpatient orders.

The primary objective of this study was to compare the quality and completeness of the DHM list using both processes. The ability to convert the DHM to an inpatient medication order was used as the primary surrogate outcome.

Methods

The nursing staff at Advocate Good Samaritan Hospital documents home medications at the time of admission. The change from a non-codified process to a codified process was part of a larger medication reconciliation and computer upgrade project. To prepare for codified home medication documentation, every nurse received 1 hour of an instructor-led class with hands-on practice. In addition, each nurse was required to complete a computer-based training module. Each of these experiences included common and complicated documentation scenarios.

The DHM lists for all inpatients for a 24 hour period just before the conversion using the non-codified process were compared with the DHM lists for all inpatients admitted during a 54 hour period following conversion to the codified process.

For the evaluation the required data elements for each DHM included: drug, dose, route of administration, frequency, and schedule. The DHM was classified as non-convertible if one or more of the required data entry fields was either missing or incorrect. Each DHM could have more than data entry deficiency.

Drug name entries were classified as non-convertible if the specific drug product could not be determined. Non-convertible drug entries were stratified into four groups: multiple salt forms, multiple dosage forms, multiple dosage strengths, and unknown. Misspellings of drug names using the non-codified process were not classified as non-convertible if the drug could be identified by the pharmacist.

Drug doses were classified as non-convertible if the data were missing, the entry was a dosing range, or was incorrect for the specific drug. Routes of administration were classified as non-convertible if the data were missing or incorrect for the specific drug.

Drug schedules were classified as non-convertible if a schedule or the date/time of the next dose was missing for drugs administered with frequencies greater than every 24 hours (ie, every other day, every 72 hours, every week, etc).

The DHM entries were classified as exceptions if the pharmacist could make some assumptions about the missing data and the entry could be converted to an inpatient order using the available data. Examples of assumptions would be a missing route of administration for obvious oral medications, a missing dose if only one dose is available for that drug, a missing dose if the dose strength was included in the drug entry, a missing dose if a dosing range was documented, and a missing frequency if a frequency range was documented.

Categorical variables were compared using the χ² test or Fisher's exact test and continuous variables were compared using the Student's t test for unpaired samples. A p value of less than 0.05 was considered statistically significant.

Results

There were 176 patients with 1618 home medications in the non-codified group compared with 94 patients with 646 home medications in the codified group, with the non-codified group having a statistically greater mean number of home medications (see table 1). There was no relationship between the number of DHM per patient and the number of non-convertible DHM entries (r²=0.18 for the non-codified group and r²=0.16 for the codified group).

Table 1.

Group comparison

Parameter	Non-codified medication process	Codified medication process	p Value
Patients	176	94
DHM	1618	646
Per patient (mean±SD)	8.2±6.4	6.3±4.8	p=0.012
Non-convertible DHM	318 (20%)	82 (13%)	p=0.0009
Drug	65 (4%)	9 (1.4%)	p=0.002
Dose	139 (8.6%)	55 (8.5%)	p=NS
Route	38 (2.3%)	25 (3.9%)	p=0.054
Frequency	124 (7.7%)	22 (3.4%)	p=0.0004
Schedule	14 (1.1%)	5 (0.8%)	p=NS

DHM, documented home medication.

All DHM were complete and could be converted to inpatient orders for 65 (37%) of the patients using the non-codified medication process compared with 53 (56%) using the codified medication process (p=0.059).

Drug identification

Non-convertible DHM due to the identification of the drug was significantly different between the two groups; 4% for non-codified entries compared with 1.4% for codified entries (p=0.002). The reasons for being non-convertible also differed (see table 2).

Table 2.

Non-convertible DHM due to the identification of the drug

Reason	Non-codified drug entries	Codified drug entries
Multiple salt forms	24 (37%)	0
Multiple dosage forms	21 (32%)	6 (67%)
Multiple dosage strengths	4 (6.2%)	2 (22%)
Unable to identify	16 (25%)	1 (11%)

DHM, documented home medication.

Non-convertible non-codified drug entries due to multiple salt forms included: potassium (6), calcium (5), iron (5), isosorbide (5), and one each for lithium, magnesium, and zinc.

Drug entries were classified as non-convertible due to multiple dosage forms if the drug entry did not specify the dosage form and the drug frequency was inconsistent with the selected drug entry. In all cases the non-convertible drugs were available in both immediate and extended release dosage forms. Examples for non-codified drug entries included metformin (4), valproic acid (4), venlafaxine (3), metoprolol (3), dextroamphetamine (2), bupropion (2), and one each for zolpidem, verapamil, and theophylline. Examples for the codified drug entries include diltiazem (2) and one each for metoprolol, morphine, guaifenesin, and venlafaxine.

The number of non-convertible drug entries due to multiple dosage strengths was small in both groups. For non-codified drug entries the drugs were ophthalmics (2) and topical (1) and for codified drug entries the drugs were ophthalmics (1) and topical (1).

Non-codified drug entries that could not be identified represent 1% of the total non-convertible DHM compared with 0.15% for codified drug entries (p=0.07). The reasons for lack of identification for non-codified drug entries included entries in which multiple drug products are available given the nomenclature used (n=7), unknown drug based upon the spelling (n=5), drug entries with a dose that is inconsistent with the drug (n=2), and drug entries representing either a combination or single ingredient product (n=2) (see table 3). The single codified drug entry that could not be identified was for ketoprofen topical, which is not commercially available.

Table 3.

Non-codified drug entries that could not be identified

Drug entry*	Reason
Calcium/magn/zinc	Multiple drugs available
Epoetin alfa 150 mcg/0.3 ml	Epoetin alfa versus darbepoetin alfa
Hydrochlorothiazide 37.5/25	Combination versus single ingredient drug
Hydrochlorothiazide-triamterene 25 mg	Combination versus single ingredient drug
Hydrocortisone enema	Multiple drugs available
Isosorbide dinitrate 30 mg daily	Dinitrate versus mononitrate drug
Lafibra	Unknown drug
MVI with calcium	Multiple drugs available
Occumin	Unknown drug
Quetrone 4 g	Cholestyramine versus unknown drug
Slow-release iron	Multiple drugs available
Stool softner	Multiple drugs available
Trimeriten	Triamterene versus unknown drug
Vita plus E	Multiple drugs available
Vita B complex	Multiple drugs available
Walgreens Waldry	Unknown drug

^*Actual drug entry including capitalization and spelling.

DHM, documented home medication.

In addition to the above there were a further 63 non-codified drug entries (4% of total DHM entries) in which the drug name was misspelled and the identification of the drug was assumed by the pharmacist.

Drug dose

There were no significant differences in the number of non-convertible drug entries due to the drug dose. There was a significant difference in the reasons for not being able to convert the drug dose entry (see table 4).

Table 4.

Non-convertible DHM due to drug dose

Reasons	Non-codified dose entries	Codified dose entries
Missing	97 (70%)	53 (96%)
Incorrect dose entered	29 (21%)	2 (3.6%)
Range dose entered	11 (7.9%)	–
Questionable entry	2 (1.4%)	–

DHM, documented home medication.

For the non-codified drug entries the dose was entered with the wrong units (ie, mcg instead of mg, mg instead of mcg, etc) in 14 cases, the dose was inconsistent with the available dosage forms (ie, 20 mg dose when 200 mg is the smallest dose form available, 30 mg dose when 75 mg is the smallest dosage form available, etc) in nine cases, the dose was inconsistent with the drug (ie, enoxaparin 1 mg, epoetin alfa 150 mcg) in five cases, and one case in which the dose was inconsistent with the frequency (alendronate 5 mg weekly).

Twenty-nine (55%) of the codified drug names were associated with a specific dosage strength. The dose was assumed to be the dosage strength during the evaluation of convertibility. For codified dose entries the incorrect dose entries included atenolol 10 mg oral daily and ranitidine 75 mg tablet 50 mg oral twice a day.

Route of administration

There was a significant increase in the number of non-convertible drug entries due to the route of administration with the codified entries, 3.9% compared with 2.3% for the non-codified entries. Twenty-four (96%) of the codified entries would be re-classified as exceptions because the oral route of administration could be inferred from the drug, dose, and frequency compared with 32 (87%) of the non-codified entries. Three of the non-convertible non-codified drug entries were intravenous compared with oral administration, one was oral compared with rectal administration, and one was intravenous compared with subcutaneous administration. The single non-convertible codified drug entry was for vitamin E without a dose, route, or frequency.

Dosing frequency

There was a significant difference between the two groups in the number of drug entries with incomplete, missing, or range frequencies, but the distribution of the reasons was similar (see table 5).

Table 5.

Non-convertible DHM due to drug frequency

Reasons	Non-codified dose entries	Codified dose entries
Incomplete	77 (63%)	11 (50%)
Missing	44 (36%)	10 (46%)
Range frequency entered	2 (1.6%)	1 (4.5%)

DHM, documented home medication.

Classification with exceptions

Twenty-eight (8.8%) of the non-convertible non-codified DHM could be re-classified as convertible based upon exceptions. Eighteen for route entries with 15 orders in which an oral route could be assumed based upon the drug and three orders in which an incorrect route was documented but the correct route could be assumed based upon the drug. Four frequency entries with three orders documented with frequency ranges and one order with no frequency documented but a standard frequency is known for the drug (nitroglycerin sublinqual). There were six dose entries with dose ranges documented.

Thirty-one (38%) of the non-convertible codified DHM entries could be re-classified based upon exceptions. Twenty dose entries, all with a dose strength included in the drug name. Six route entries, all in which an oral route could be assumed based upon the drug. Five dose/route entries, all in which an oral route could be assumed based upon the drug and the dose strength was included in the drug name.

When exceptions are considered, the use of the codified process resulted in a statistically significant increase in the number of complete drug entries (92% vs 82%, p<0.001) and in the number of patients with a complete DHM list (70% vs 42%, p=0.014) when compared with the non-codified process.

Discussion

A DHM list containing all of the medication order data elements helps to ensure continuity of patient care, prevent medication errors, and provide a medication history from which therapy adjustments can be made. Incomplete or inaccurate DHM entries result in additional work for the pharmacy and nursing staff^{9 10 12} and increase the likelihood of a medication error occurrence.^{3 7 12} If a reconciliation of this information is not done; incorrect dosing, route of administration, and frequency information may be used for an inpatient medication order.^{3 7 12}

The free-text nature of the non-codified data entry allows for inaccurate information to be entered, whereas the codified process helps to prevent these types of errors. The accuracy of the information with a non-codified process is dependent upon the multiple checks and balances of the inpatient medication ordering procedures to prevent these types of medication errors, but they still occur.¹⁹

The conversion of DHM lists from non-coded to coded data has been described. In one study,⁵ non-coded narrative medication lists were converted to coded data using a natural language process. The results show that conversion of non-coded data is hampered by the use of abbreviations and other non-medical terminology in the dataset. The present study confirms the limitation of this approach, with 11% of the non-coded DHM having at least one of the four required data fields (drug, dose, route, and frequency) containing inaccurate or incomplete information compared with 1.5% of coded DHM.

The integration of the codified DHM list with the inpatient computerized provider order entry system enabled the conversion of DHM orders to inpatient orders. As cautioned by others,²⁰ reliance on the data provided by the knowledge-driven medication order functionality created some unintended medication errors. In the present study, despite the rigorous nature of the functionality, three DHM entries were incorrect. The nature of these errors would have prompted an intervention by either the nurse or the pharmacist before an attempt to convert to an inpatient. However, there were a further five DHM entries in which if converted, the patient would have received the incorrect dosage form of the drug.

Similar to other systems^{5 6 13 17} the upgraded electronic medication record system enabled the use of the patient's previous electronic discharge medication list as a starting point for the codified DHM list. Following the implementation of an electronic DHM list, using data converted from a free-text form, one hospital experienced a significant decrease in unreconciled medications from 15% to less than 5% when compared with the use of a hand-written DHM list.¹⁸ Another hospital, following a similar implementation, reported a reduction in admission medication reconciliation interventions related to the medication order from 18% to 17%.⁴ Whether the improvement was because of the change in the completeness of the medication orders or an improvement in the overall medication reconciliation process was not evaluated. As the present study was conducted soon after implementation of the codified DHM process, the impact of the conversion capability on the entire medication reconciliation process, including admission and discharge, was not evaluated. The results therefore apply to only the first step in the process and an evaluation of the impact on discharge medication reconciliation is warranted.

A limitation of the study is the time period from which the non-codified and codified groups were collected. The non-codified group represented a patient population (ie, the current inpatient census) for which the DHM lists were readily retrievable without accessing archived data. Data from the present group are representative of the non-codified process because it encompasses all patient care units, a significant cross-section of the nursing staff, and DHM lists recorded over a period of several days to weeks before the implementation of the new process. It should not differ significantly from any other non-codified patient group because there had been no change in the non-codified process for several years. The time period for the codified group was determined by the need for a rapid evaluation of the new codified process and the educational programme. Results from this group represent a worst case scenario because, with additional experience, the nursing staff would have become more proficient with the new process.

The training of the nursing staff took place immediately before the study period, which may have introduced some bias; however, any bias would have been in favor of the non-codified group results. Due to the significant differences in the processes, that is free-text entry compared with computer-driven menus, it is unlikely that any lessons learned during the education programme could have been applied to the non-codified process.

The present study did not attempt to evaluate the accuracy of the DHM against another source of truth; the patient or another medical record source. It is possible that some of the inaccurate drug and drug dosing information was based on an inaccurate source, but this should not have affected the overall results because the source was similar in both groups.

The study did not include an evaluation of the time needed by the nurse to enter either the non-codified or codified DHM because the implementation of the codified process was mandated by the healthcare system informatics department. Subjectively, data entry using the codified process was significantly longer. The nurse was required to navigate through several different screens to enter one codified DHM, whereas the free-text nature of the non-codified process could be accomplished using one data entry screen for all DHM. One advantage of using the codified process is the ability to generate a discharge medication list from the DHM. Previously, this list was handwritten by the nursing staff. Whether the time needed to enter the codified DHM list was offset by the time saved in creating the discharge medication list was not evaluated.

The study also did not evaluate differences in accuracy or efficiency in data entry between nurses, pharmacists, and/or physicians as the DHM process is a nursing responsibility for this hospital. The codified DHM process relies heavily on the use of drug databases used by pharmacists, knowledge of both trade and generic names of drugs, and familiarity with common drug dosage strengths and dosage forms. Data entry by pharmacists would probably be more accurate and efficient. A comparison of data entry between nursing and pharmacy staff would be an interesting area for future research.

Conclusion

The methodology used to document home medications can have a significant impact on patient safety. The use of a codified medication process to document home medications results in a more complete and accurate list of medications, resulting in less potential for patient harm and less reliance on the medication ordering process to prevent inadvertent medication errors.