INTRODUCTION

Concerns about the quality= of services offered in healthcare facilities have been widely discussed, mainly due to the alarming rates of adverse events that cause harm to patients at different healthcare levels⁽^1-2)=. There is a contemporary trend in a worldwide discussion on patient safety. = In Brazil, public healthcare is provided by the Unified Health System (SUS in Portuguese), which offers universal access to integral care by means of a national health. There is an increasing concern about patient safety and ca= re quality in facilities in the SUS system due to society's dissatisfaction wi= th bad practices of care offered in such facilities^(<=
/sup>^3-5). Despite the widespread dissemination of patient safety studies developed in= the hospital environment, it is known that there are gaps regarding this theme = in primary healthcare centers or facilities⁽^4,6).

Medication errors affect care in a very dangerous = way, generate unnecessary costs, extend the duration of treatments and the patie= nts' length of stay, and can lead to litigious actions⁽⁼^7-8). This can result in discrediting of the professionals' performance and of the institutions involved. It is also known that safe behaviors are associated = with the reduction of adverse events⁽^=
2,9-¹⁰⁾.

In view of the relevance of the theme, this study aimed to identify errors in drug prescriptions and its causes in a primary healthcare center, taking into account the country's legal provisions regarding drug prescription.

MATERIAL AND METHODS

A cross-sectional study was developed in a primary healthcare center in northeastern Brazil. Patients w= ho attend to this center receive medications to use at home. The medications a= re dispensed in the center's pharmacy for patients carrying prescriptions orde= red by a health professional (nurse, physician, or dentist) from the center. A = copy of each prescription is retained in the center's archive. This study analyz= ed prescriptions retained in the center.

A retrospective analysis of 707 drug prescriptions was carried out. These prescriptions were dispensed = from August to November 2017. A form was developed and used for analyzing the content of the prescriptions, with pre-established criteria from the legal provisions of the Brazilian Decree No. 74.170/74, Ordinance No. 344/98, and Resolution No. 357/01. The main regulations include the following: (a) being written in Portuguese and in ink; (b) being legible (the prescription is readable, easy to understand, and does not lead to a dispensing error); (c) inclusion of the patient's name and address; (d) use of the official nomenclature of the drug or trade name; and (e) being complete, i.e., must include: the pharmaceutical form, drug dosage, presentation and method of administration, the duration of treatment, prescription's date of issue and validity, and the prescriber's name, registration number, signature and sta= mp. The presence of erasures or amendments was also observed.

The form had questions abo= ut prescription quality indicators, such as: data about the patient (sex and a= ge), prescriber profession, date of prescription, legible or illegible handwriti= ng and, finally, general information about the drugs (name, quantity, therapeu= tic class, drug administration route, and time and frequency of administration). The prescriptions were read and analyzed by three researchers, separately, = and after their interpretation the data were compared for a more detailed assessment.

Data were computed in a= Excel spreadsheet and then analyzed with the Epi Info™ statistical software 7.2.1= .0 version (CDC, Atlanta, USA). Descriptive and inferential statistical were u= sed to analyze the collected data. The Pearson's chi-square was used to verify the existence of link among the categorical variables. A Kolmogorov-Smirnov normality test was performed to determine if the data were normally distributed. Considering the non-parametric nature of the variables, the Mann-Whitney test was used to compare the ranks of quantities of prescribed drugs. A 0.05 significance level was stabilized for all analyses.

<= /p>

Ishikawa diagram

An Ishika= wa diagram was created to identify the “root causes” for the identified errors= . A root cause is a fundamental reason for the occurrence of a problem. This me= thod is proposed not only to find root causes, but to find solutions and to prev= ent new episodes (other errors) from happening. The diagram is a graphic representation that illustrates the relations between a specific result and= its causes.

In the present study, the Ishikawa diagram was created to identify the main causalities that could be related = to the occurrence of errors or inadequacies in the filling of drug prescriptio= ns. The diagram was the result of a focus group meeting between the researchers= and 55 professionals from primary healthcare teams and community health agents,= to obtain better and more detailed information about the work environment and relationships. The identified causes were grouped into 6 (six) categories: personal factors, organizational fact= ors, external factors, patient factors, work/environment factors, and other fact= ors.

Ethical Procedures

The study was approved by the Research Ethics Committee of the University of International Integration of the Afro-Brazilian Lusophony - UNILAB (CAAE No. 62673716.6.0000.5576, opinion No. <= /span>2,522,957).

RESULTS

A total of 707 prescriptions were analyzed. Altogether, 1314 drugs were prescribed. The average number of drugs per prescription was 1.8. Table 1 provides an overv= iew of the variables related to the analyzed prescriptions.

Table 1 - Distribution of variables related to drug prescriptions in a Primary healthcare center, Bra= zil, 2017.

VARIABLES=	n	[%]	Confidence Interval [95%]
Gender (n =3D 667)
Male	235	[35.23]	31.70 - 38.93
Female	432	[64.77]	61.07 - 68.30
Prescriber profession (n = =3D 698)
Physician	526	[75.36]	72.03 - 78.41
Nurse	172	[24.64]	21.59 - 27.97
Num= ber of drugs per prescription (n = =3D 707)
One	300	[42.43]	38.84 - 46.11
Two	249	[35.22]	31.79 - 38.81
Three	113	[15.98]	13.47 - 18.87
Four	27	[3.82]	2.64 - 5.50
Five	14	[1.98]	1.18 - 3.30
Six	04	[0.57]	0.22 - 1.45
Therapeutic class - Prescription 1= (n= =3D 696)
Analgesic=	115	[16.52]	13.95 - 19.46
Antimicrobial	208	[29.28]	26.60 - 33.39
Anti-inflammatory	58	[8.33]	6.50 - 10.62
Antihypertensive=	148	[21.26]	18.39 - 24.46
Antiglycemic	20	[2.87]	1.87 - 4.40
Digestive Drugs<= /span>	31	[4.45]	3.16 - 6.25
Bronchodilators / Antispasm= odic	01	[0.14]	0.03 - 0.81
Anti-Cough / Flu medicines<= o:p>	02	[0.29]	0.08 - 1.04
Antianemic	19	[2.73]	1.75 - 4.22
Antiparasitic	58	[8.33]	6.50 - 10.62
Others	36	[5.17]	3.76 - 7.08
Adm= inistration route - Prescription 1 (n= =3D 681)<= /span>
Oral	651	[95.72]	93.96 - 96.99
Topic	30	[4.28]	3.01 - 6.04
Readability (n =3D 707)
Unreadable	136	[19.24]	16.50 - 22.30
Readable<= /p>	571	[80.76]	77.70 - 83.50
Information quality (n = =3D 707)
Compromised	138	[19.52]	16.77 - 22.60
Non compromised<= /span>	569	[80.48]	77.40 - 83.23
Compromised information (n = =3D 707)
Patient Name	31	[4.38]	3.11 - 6.11
Dosage	116	[16.41]	13.86 - 19.32
Route of Administration	07	[0.99]	0.48 - 2.03

Source: The authors.<= /span>

&nbs= p;

Table 2 - Factors associ= ated with prescribers' professional category in a Primary healthcare unit, Brazi= l, 2017.

Prescriber	Readability	Statistic [p-value]
Readable [%]	Illegible [%]
Prescriber
Physician=	401 [76.24]	125 [23.76]	0.000¹
Nurse	167 [97.09]	05 [2.91]
	Information quality
	Non compromised	Compromised
Prescriber
Physician=	399 [75.86]	127 [24.14]	0.000¹
Nurse	167 [97.09]	05 [2.91]
	Information: Patient's Name
	Non compromised	Compromised
Prescriber
Physician=	495 [94.11]	31 [5.89]	0.000¹
Nurse	172 [100.0]	00 [0.00]
	Information: Dosage<= /span>
	Non compromised	Compromised
Prescriber
Physician=	420 [79.85]	106 [20.15]	0.000¹
Nurse	167 [97.09]	05 [2.91]
	Information: Route of Administration
	Non compromised	Compromised
Prescriber
Physician=	519 [98.67]	07 [1.33]	0.203²
Nurse	172 [100.0]	00 [0.00]
^= 1Pearson's chi-square test; ²Fisher's exact test.

Source: The authors.

DISCUSSION

Before any reflection, it is important to understand the complex nature of human error= s. Errors are part of human nature and will always exist. However, systems in which the production, preparation and dispensing of drug prescriptions takes place must follow well-structured work routines to promote actions aimed at minimizing errors, and thus, reducing damage to the patients. Driven by an intense concern with the quality of the services offered in healthcare, many researchers have discussed what are the main barriers faced by professional= s to engage a safety culture in health services, in line with a search for tacti= cs to be used to prevent adverse events⁽= ¹¹⁾.

In this scenario, there is a need to identify the causal factors involving medicati= on prescription errors. The occurrence of these errors is directly related to = the technical quality of the prescription, mainly with its legibility and completeness⁽¹²⁾. According to t= he study findings, approximately 20% of the prescriptions still have problems = in this aspect due to not readable handwriting. Historically, prescribers' handwriting has been cited as an element that compromises the proper use medications, and consequently, patient safety⁽^13-15).

In additi= on to identifying weaknesses present in the stage of prescribing, it is essential= to adopt safety models, some in the form of assistance protocols, in order to = make the production of care and communication a systematic and equally knowledge= able act⁽¹⁶⁾. With regard to the medication process, all stages are of unparallel importance, however the prescription itself deserves special attention considering that it can be compared to a generating mechanism and guiding commands, which will lead “what”, “how”, “where”, “when”, and on “whom” interventions will be carried= out⁽^17-18).

In this study, the legibility criterion was established when observable words or expressions were written in an understandable way, allowing a clear understanding of the prescription without the need for deduction or supposition. It is known that people are more likely to use deductions when they read a prescription with a bad handwriting. This form of decoding (usi= ng purely deductive reasoning) carries with it a degree of imprecision with ha= rmful repercussions⁽¹⁹⁾. Bad handwriting and the existence of incomplete information are factors that predispose to harmful medication errors⁽^20-21).

When considering a significance level of 5%, the comparison made using the chi-square test between the two categories of prescribing professionals in relation to legibility, quality of information, and presence of the patient= 's name and dosage showed that the physicians were more likely to make mistake= s.

According= to the study findings, 138 prescriptions (approximately 20%) had some abbrevia= tion or acronym. The inappropriate use of acronyms and the non-standardization of the nomenclature of medicines also lead to difficulties in the understandin= g of prescriptions⁽¹⁴⁾. These difficulties can be associated with the massive variety of drugs currently available on the market, linked to the fact that many of them have very sim= ilar trade names⁽^16,22).

The use of abbreviations, a widely adopted practice by health professionals, can lead = to confusing or misinterpretations⁽^=
18,23). In Brazilian primary healthcare services, drugs are obtained in the same ce= nter were consultations are carried out, but patients= are free to purchase them via commercial channels. Depending on the service, the form of acquisition of medications, or even by professionals' convenience, prescriptions can be interpreted dubiously. This can be made worse if the handwriting is hard to be read⁽^1=
-2,⁴⁾.

The use of abbreviations needs to be avoided as much as possible, especially in primary healthcare services in which patients leave the center carrying out the prescriptions. An acronym may be usual for a health professional but may no= t be clear and known to the patient. Thus, professionals should avoid abbreviati= ng information about the pharmaceutical form to be used, the route of administration, doses, quantities, and intervals of int= ake⁽^8,12,24). A study carried out in primary healthcare services in Sweden, involving clinical-pharmaceutical recommendations, identified 349 problems related to medication intake caused by difficulties in identifying guidelines in prescriptions. The authors have highlighted the importance of rigor in prop= erly filling in information about the correct intake of drug= s⁽²⁵⁾.

One of the foundations of the safety culture is the adequate identification of the pat= ient at all times, and in the case of assistance provided by primary care servic= es (in which the patient takes the prescription with him to his home), this is even more important⁽^26-27). Frequently, there is more than one person in Brazilian homes being treated = for chronic or degenerative diseases. Arterial hypertension, for example, has extremely different pharmacological therapies. In the present study, the patient's name was missing in 31 prescriptions. The correct identification = of the patient is essential for the implementation of a duly safe assistance⁽^14-15,27-28).

Ishikawa diagram

Regarding the Ishikawa diagram, the Personal factors category included the main causes that compromised patient safety, such as personal training, excess of self-confidence, haste/automatism, and finally, insufficient knowledge. Professional training is an attitude to be performed by everyone who enters= a specific type of service, combined with permanent educa= tion⁽¹⁴⁾. Self-confidence combined with automatism in the performance of functions enhances the probability of errors⁽29).

In the foll= owing category, Organizational factors, we pointed out the lack of supervision, failures in communication (illegible handwriting), and the absence of proto= cols or flowcharts for prescription. For a safe practice, it is necessary to use= and implement protocols that support the work routine⁽⁼^19,29-30).

In the third category, External factors, the absence of a permanent evaluation policy and the containment of expenses were evidenced. External evaluation and monitor= ing policy are important tools for maintaining a minimum acceptable standard⁽^15-16). This policy must involve analyzes of compliance with conducts.

In category= No. 4, Patient factors, medication culture and hastiness were listed. The fifth category, Work/environment factors, workload, the absence of auditing, and = the inadequacy of the size of the pharmacy room were the root causes for the prescribing errors⁽³⁰⁾. The existence of quality standards in the provision of care depends on periodic inspections and an audit policy⁽^=
12). Clinical audits, for example, work as a systematic analysis of clinical procedures aimed at improving the quality and results of care. In addition,= the existence of regulatory inspections is important to control and verify procedures, observing if they are in compliance with all legal requirements⁽^9,12,31). Professio= nals must understand the importance of audits and the adherence to protocols as a way to ensure patient safety and quality of care^(<=
/sup>^9,18).

Limitations=

The main limitation of this study concerns the relati= vely small number of healthcare workers who were able to participate. Some professionals were interested in participate but were not available due to their busy schedule and this factor, particularly for physicians, might have limited our findings. This study is also limited by the use of only one pri= mary healthcare unit.

Implication= s

Through the= use of the Ishikawa diagram we were able to ascertain w= hich were the most likely causal factors for the occurrence of prescription erro= rs at a primary healthcare center, and also in the identification of needs for improvements in management, professional training/improvement, use of resources, and ongoing training of the health teams.

CONCLUSIONS

The study made it possible= to assess that the main errors involving drug prescriptions in primary healthc= are were related to handwrite illegibility and lack of information such as drug dosage and patients' name, reflecting on the quality of the information ent= ered in the prescriptions. The physicians' category was associated with a greater occurrence of errors.

The construction of the Ishikawa diagram allowed the identification of root causes of the observed errors. They were more directly related to organizational factors, personal factors and service-related factors. However, the results must be analyzed = with caution, since this is a cross-sectional study, carried out in a predetermi= ned period and, therefore, cannot be universally representative.

This study provides subsid= ies for future research. New studies should be carried out aiming at a better understanding of which factors (including professional training and permane= nt training) contribute to the effectiveness of drug prescription practices and the accuracy of information. Findings from this study can be used to set go= als to improve the quality of services in relation to prescribing errors, provi= ding critical reflection and, consequently, strengthening the public health syst= em.

REFERENCES<= /p>

1. Álvares J, Alves MCGP, Escuder MML, Almeida AM, Izidoro JB, Guerra Junior AA, et al. <= /span>National survey on access, use and promotion of rational use of medicines: Methods. Rev Saúde Públ. 2017;51:1s-9s. doi: 10.11606/S1518-8787.2017051007027

2. Vilela RPB, Jericó MC. Implementing technologies to= prevent medication errors at a high-complexity hospital: analysis of cost and resul= ts. Einstein (São Paulo). 2019;17(4):eGS4621. doi: 10.31744/einstein_journal/2019gs4621

3. Melo DO, Silva SRA, Castro LLC. Evaluation of quality indicators = of prescription drugs in a primary care unit with different models of care. Rev Epidemiol Serv Health. 2016;25(2):259–70. doi: 10.5123/S1679-49742016000200005

4. Santos LDM, Falk JA. Electronic prescription syst= em in a home care company: perception of doctors and nurses. Rev adm saúde. 2019;19(75). doi: 10.23973/ras.75.16= 4

5. Gala P, Moshokgo V, Seth B, Ramasuana K, Kazadi E, M’buse R, et a= l. Medication Errors and Blood Pressure Control Among Patients Managed for Hypertension in Public Ambulatory Care Clinics in Botswana. J Am Heart Assoc. 2020;9(2). doi: 10.1161/JAHA.119.013766

6. Vasconcelos PF, Freitas CH, Jorge MS, Carvalho RE, Freire VE, Araújo MM, et al. Safety attributes in primary care: understanding the needs of patients, health professionals, and managers. Public health, 2019;171:31-40.

7. Ministry of Health (BR). Reference document for= the National Patient Safety Program. Ministry of Health; 2014.

8. Costa DB, Macedo LLA, Souto RADM, Santos AL. Drug Prescription Errors: Evaluation of prescriptions in the pediatrics clinic o= f a school hospital. Rev Bras Farm Hosp Serv Saúde. 2019;9(2):1–5. doi: 10.30968/rbfhss.2018.092.002

9. Marchon SG, Mendes Junior WV. Patient safety in primary health ca= re: a systematic review. Cad Saúde P= ública. 2014;30(9):1815–35. doi: 10.1590/0102-311X00114113

10. Santos PRA, Rocha FLR, Sampaio CSJC. Actions for safe= ty in the prescription, use and administration of medicines in emergency care uni= ts. Rev gaúch enferm. 2019;40(spe):e20180347. doi: 10.1590/1983-1447.= 2019.20180347

11. Kohn LT, Corrigan JM, Donaldson MS. To Err Is Human. Building a Safer Health System, Volume 6. J Public Health. 2000;2(3):93–5. doi: 10.17226/9728

12. Coetzee R, Johnson Y, van Niekerk J, Namane M. Amitriptyline prescribing in public sector healthcare facilities in the Wes= tern Cape, South Africa. PLoS ONE. 2020;15(4). doi: 10.1371/journal.pone.0231675

13. Gates PJ, Baysari MT, Mumford V, Raban MZ, Westbrook JI. Standardising the Classification of Harm Associated with Medication Errors:= The Harm Associated with Medication Error Classification (HAMEC). Drug Saf. 2019;42(8):931–9. doi: 10.1007/s40264-019-00823-4

14. Ting HW, Chung SL, Chen CF, Chiu HY, Hsieh YW. A drug identification model developed using deep learning technologies: experience= of a medical center in Taiwan. BMC Health Serv Res. 2020;20(1):312. doi: 10.1186/s12913-020-05166-w

15. Gogazeh E. Dispensing errors and self-medication practice observed by commu= nity pharmacists in Jordan. Saudi pharmaceutical journal: SPJ. 2020;28(3):233–7. doi: 10.1016/j.jsps.2020.01.001

16. Figueire= do TWB, Silva LAA, Brusamarello T, Oliveira ES, Santos T, Pontes L. Tipos, Cau= sas E Estratégias De Intervenção Frente a Erros De Medicação: Uma Revisão Integrativa. Rev enferm atenção saúde. 2018;= 7(2):400. doi: 10.18554/reas.v7i2.2494

17. Larmené-Beld KHM, Alting EK, Taxis K. A systematic literature re= view on strategies to avoid look-alike errors of labels. Eur J Clin Pharmacol. 2018;74(8):985–93. doi: 10.1007/s00228-018-2471-z

18. Lizano-Díez I, Figueiredo-Escri= bá C, Piñero-López MÁ, Lastra CF, Mariño EL, Modamio P. Prevention strategies to identify LASA er= rors: Building and sustaining a culture of patient safety. BMC Health Serv Res. 2020;20(1). doi: 10.1186/s12913-020-49= 22-3

19. Schrader T, Tetzlaff L, Beck E, Mindt S, Geiss F, Hauser K, et al. The similarity of drug names as a possible cause of confusion: Analy= sis of data from outpatient care. Z = Evid Fortbild Qual Gesundhwes. 2020. doi: 10.1016/j.zefq.2020.01.006<= /o:p>

20. Islam MS, Faisal T, Munira MS, Ahammed MS, Saha S, Afrin S. Poor Understanding of Prescribed Drugs Leads to Medication Error at Both Pharmacy and Patient End: A Survey Study to find Out the Underlying Factors. Medical Research and Clinical Case Rep= orts. 2018;2(4):260–72.

21. Manias E. Effects of interdisciplinary collaboration in hospital= s on medication errors: an integrative review. = E= xpert Opin. Drug Saf. 2018;17(3):259–75. doi: 10.1080/14740338.2018.1424830

22. Ministry of Health (BR). How can I contribute = to increasing patient safety? Natio= nal Health Surveillance Agency; 2017.

23. Tomlin AM, Woods DJ, Lloyd HS, Tilyard MW. Trends in Outpatient Prescription Medicine Use in New Zealand Children 2010-2015: A National Population-Based Study. Paediatr drugs. 2018;20(5):465-74. doi: 10.1007/s40272-018-0303-3

24. Suclupe S, Martinez-Zapata MJ, Mancebo J, Font-Vaquer A, Castillo-Masa AM, Viñolas I, et al. Medication errors in prescription and administration in critically ill patients. J Adv Nurs. 2020;76(5):1192–200. doi: 10.1111/jan.14322

25. Modig S, Holmdahl L, Bondesson Å. Medication reviews in primary = care in Sweden: importance of clinical pharmacists’ recommendations on drug-rela= ted problems. Int J Clin Pharm. 2016;38(1):41–5. doi: 10.1007/s11096-015-0189-x

26. Curran C, Lydon S, Kelly M, Murphy A, Walsh C, OʼConnor P. A Systematic Review of Primary = Care Safety Climate Survey Instruments: Their Origins, Psychometric Properties, Quality, and Usage. J Patient Saf. 2018;14(2):e9= –e18. doi: 10.1097/PTS.0000000000000393

27. Lockwood AM, Proulx J, Hill M, Pendray J. Using safety culture results to guide the merger of four general practices in the UK. BMJ Open. 2020;9:e000860. doi: 10.1136/bmjoq-2019-00= 0860

28. Franklin BJ, Gandhi TK, Bates DW, Huancahuari N, Morris CA, Pear= son M, et al. Impact of multidisciplinary team huddles on patient safety: A systematic review and proposed taxonomy. BMJ Qual Saf. 2020. doi: 10.1136/bmjqs-2019-009911

29. Teixeira TCA, Cassiani SHB. Root cause analysis of falling accid= ents and medication errors in hospital. A= cta Paul Enferm. 2014;27(2):100-7. doi: 10.1590/1982-0194201400019

30. Allen SM, Kachlic MD, Parent-Stevens L. Pharmacy Students Teaching Prescription Writing and Nonprescription Product Selection to Medical Students. Am J Pharm Educ. 2020;84(3):6972. doi: 10.5688/ajpe69= 72=

31. Schiesari LMC. External evaluation of hospital organizat= ions in Brazil: can we do it differently? Ciênc Saúde Colet. 2014;= 19(10):4229–34. doi: 10.1590/1413-812320141910.2164201

<= /p>

Corresponding author: Nayara Crist= ina Rabelo Bandeira Address: Betel street, no number – Dendê, Fortaleza, Ceará, Brazil. Telephone number: (85) 98504 9728 E-mail: nayarabandeira@gmail.com<= /o:p>

Submission: 2021-05-25

Approval: 2021-07-28

<= /p>

GROUPS	Quantity of prescribed drugs	Statistic [p-value]¹
n	Average	[SD]	Median	Mean Rank
Prescriber (n =3D 698)
Physician	526	1.93	[1.01]	2.0	359.43	0.015
Nurse	172	1.72	[0.89]	2.0	319.14
Patients' Gender (n =3D 667)
Male	235	1.87	[1.00]	2.0	332.82	0.901
Feminine	432	1.88	[0.98]	2.0	334.64