Summary of main findings
The frequency of medical errors found in the medical records in the 12 clinics in Malaysia in this cross-sectional study of 1753 records was: documentation errors 98.0%; medication errors 41.1%, investigation errors 21.7%, decision making errors 14.5% and diagnostic errors 3.6%. The majority of these patient encounters were with medical assistants. Almost all medical records had some form of documentation problem. Diagnostic and management errors were found in half of the patient records. Most errors were considered to be preventable and 40% of the errors were viewed as having a potential for causing serious harm.
This is the first study conducted on medical errors in primary care settings in Malaysia. Other studies on medical errors in primary care, used incident reporting [2, 11, 12] and examined the different aspects of errors [5, 13–15]. We found the rate of medical errors in primary care to be unacceptably high with 40% of the errors having the potential to cause serious harm.
We found an extremely high rate of documentation error. This is in contrast to a study in the US where clinician self-reported missing information per patient visit was only 13.6% . In developed countries, legibility may not be an issue because of electronic records. The primary care clinics in this study were not computerized. Another contributing factor to poor documentation could be the heavy workload at the clinics, which ranged from 1 to 13 patients per personnel per hour in participating clinics.
We found the medication error to be 41.1%. Other studies have reported a wide range of error rates, ranging from less than 1% to 52% per 100 incidents reported [5, 7, 8, 11, 12, 17, 18]. Some studies reported treatment errors, which could be either medication errors or errors in other forms of treatments, that ranged between 15% and 47% per 100 incident/errors reported [2, 6, 12, 15, 17, 18]. Common medication errors found included simultaneous prescription of two antihistamines for upper respiratory tract infections (for example, one for cough and one for nasal decongestant) and prescription of antacids with non-steroidal anti-inflammatory drugs (NSAID) for perceived prevention of NSAID-induced ulcer.
The occurrence rate of investigation errors was 21.7%, which is comparable with findings from studies done in the US and Australia [2, 17]. However, when one takes into account the inconclusive errors, the occurrence rate could potentially be higher. Our definition of investigation errors included investigations that should have been done but were not ordered to be carried out. Resource constraints for investigations in some clinics are a possible contributing factor.
Decision making errors
There were 14.5% of the records that had decision making errors. However, 40.2% of the records were inconclusive for these errors. The main error noted was inappropriate period for follow-up care. Inappropriate referrals (where referrals were inappropriately given or omitted) occurred in 6.3% of the consultations. We were unable to differentiate whether this was due to lack of awareness and knowledge or judgment errors. In some studies, judgment error was reported to range from 2% in the US to 44% in Australia [11, 12, 17, 18].
Although we found a low rate of diagnostic errors (3.6%), in 61.9% of the records, the diagnosis was deemed to be inconclusive due to inadequate documentation, likely causing an underestimation of errors. A number of records had only the presenting problem or diagnosis written, with no documentation of history or physical examination. The actual rate of error is likely to be much higher because of such high inconclusive rates. Other studies have found the percentage of diagnostic errors out of total potential errors to range between 1.3% and 78% [2, 5, 7, 8, 11, 17, 19–21]. Apart from poor documentation, diagnostic errors could be attributed to a lack of knowledge and expertise of the health care providers. Most (81%) of the patient encounters were managed by medical assistants who provided a wide scope of medical services at a level similar to nurse practitioners in primary care clinics. Their service is important especially in the medically under-served areas. It is therefore important to provide continuing medical education to this group of practitioners.
Preventability and likelihood to cause harm
About 40% of the errors were assessed by the expert panel of family medicine specialists as having the potential to cause serious harm, either in the short or long term. This is consistent with the findings of other studies where potential harm was deemed likely to occur in 5.8% to 49% [15, 17]. Most of the errors were perceived to have strong evidence for preventability, which again was consistent with findings from other studies [5, 7, 12, 22]. In view of the high rate of potential serious harm, it is imperative for urgent measures to be taken to reduce the rate of medical errors in primary care.
Currently in Malaysia, there is lack of a formal reporting mechanism for medical errors in primary care. Incident reporting is practiced as a self-reporting process, and the number of incidents reported is influenced by a “blame-free” culture. Internationally, only 30% of incidents are said to be reported [5, 11]. A workable system is needed to detect and investigate errors in primary care.
Implications for clinical practice and future research
The occurrence of medical errors in primary care was high and consistent with international literature. This study found extremely high rates of documentation errors and this should be the area of greatest priority for intervention. Interventions targeting legibility of handwriting or structured or electronic record keeping are possible system changes that can be considered. Medication errors can be reduced by the use of a tool or drug formulary that summarizes information on commonly used drugs, their dosages and frequencies for use.
The likelihood of errors to possibly cause serious harm was high, of which most were preventable. A concerted effort is needed to improve patient safety in primary care that includes system changes, continuous education, monitoring of implementation and policy change.
The findings of this study on medical errors call for more studies of this nature as well as interventional studies to reduce medical errors. Further research is needed to identify organizational determinants and latent failures to improve the system. A review of interventions that were successful elsewhere is needed and their applicability to the local context is required. In addition, the magnitude of errors in general practice in the private sector is unknown and should be studied.
Strengths and limitations of the study
Selection of the clinics was by purposive sampling. An equal number of clinics, supervised by different levels of expertise, were selected from two different geographical regions to increase the validity of this study. In addition, the medical records were selected randomly.
We assessed errors by reviewing medical records, and not on errors reported by practitioners. This method of error detection would give a better reflection of the extent of medical errors as opportunistic incidents reporting tend to underestimate errors .
We were not able to detect all adverse events due to misdiagnosis, delayed diagnosis or missed diagnosis as this was a cross-sectional study. Therefore, diagnostic errors could have been underestimated. We also did not study patient factors that could contribute to errors in management such as patient’s request for investigations, refusals for referral or for admissions, which could affect the rate of decision making error. Furthermore, the majority of patient encounters in the medical records assessed were with medical assistants, with only a small number of encounters with medical officers and family medicine specialists.
Poor documentation was a major problem in this study, with the possibility that the magnitude of errors was underestimated. This is however a very important finding which requires immediate remedial action. The medico-legal implications are of concern.
This study aimed to identify areas of errors that have occurred. We did not examine the root causes of errors from which a definitive intervention could be instituted. Further research should study the causes of errors and investigate possible corrective interventions.