Incident reporting in nurse-led national telephone triage in Sweden: the reported errors reveal a pattern that needs to be broken

Abstract

We conducted a retrospective study of incident reports concerning the national, nurse-led telephone triage system in Sweden. The Swedish Health Care Direct organization (SHD) is staffed by registered nurses who act as telenurses and triage the callers' need for care, using a computerized decision support system. Data were collected during 2007 from all county councils that participated in the SHD and were analysed using content analysis. Incident reports were then compared concerning differences in reported categories and who reported the errors. The 426 incident reports included 452 errors. Of the analysed incident reports, 41% concerned accessibility problems, 25% incorrect assessment, 15% routines/guidelines, 13% technical problems and 6% information and communication. The most frequent outgoing incident reports (i.e. sent from SHD to other health-care providers) concerned accessibility problems and the most frequently incoming reports (i.e. sent to SHD from other health-care providers) concerned incorrect assessment. There was a significant difference (P < 0.001) between outgoing and incoming reports regarding the main category. Telenurses have limited possibilities for referring the caller to their primary health-care provider or specialist, which may cause them to over-triage or under-triage the callers' need for care. This over-triage or under-triage may in turn cause other health-care providers to report incorrect assessment to SHD. The implications for practice are that poor accessibility is a matter that should be addressed and that the reasons for incorrect assessment should be explored.

Introduction

In Sweden, a centralization of telenursing started in 2003 with the implementation of a national telephone helpline called Swedish Health Care Direct (SHD). The service is staffed by registered nurses who triage the callers' need for further care using a computerized decision support system (CDSS).¹ Possible outcomes of a call include self-care advice, an appointment with a General Practitioner (GP), a visit to the accident and emergency department or a request for an ambulance.^1,2

Telenurse work is complex and knowledge-intensive. The telenurses have numerous patient encounters every day, and calls include a broad range of medical conditions and age groups. Telenurses work independently and make decisions about the need for further care, give self-care advice or refer the caller to another caregiver.³ When telenurses triage callers, their assessments are based on telephone communication only and this places great demand on their communicative skills and ability to listen.⁴

Previous studies have shown that if staff always have to make decisions, the experience is stressful and demanding.^4,5 The call-centre culture demands efficiency in making assessments without seeing the patient. The callers have different capacities to communicate their problems, and telenurses have reported that callers sometimes exaggerate or understate their problems.⁴ This, combined with the broad range of problems presented to the telenurses, may result in errors.

To date, no studies have dealt with telenursing and incident reporting, except one that compared the number of adverse events before and after a telenursing trial.⁶ The aim of the present study was to describe errors that lead to an incident report within the context of SHD telenursing. The research questions were:

What errors are reported in the incident reports?

Who reports what type of error?

Are there any differences between incoming incident reports (i.e. those sent to SHD from other health-care providers or patients) and outgoing ones (i.e. those sent from SHD to other health-care providers)?

Methods

The study was approved by the appropriate ethics committee.

Six of 21 county councils in Sweden were connected to SHD in 2007. All incident reports during 2007 were collected from five of these sites. The sixth site had a high frequency of reported incidents and the director of operations was instructed to select slightly more than 20% of all incident reports, stored in manual files. Thus, a total of 426 reports were collected, and based on these reports 452 errors were identified (i.e. one incident report could contain more than one error). Five sites used paper-based incident reporting schemes and one used computer-based schemes. All incident reports in Sweden are non-confidential. The analysed incident report forms had the same content but different layout (see Table 1).

Table 1

Items on the incident report form

Headings	Information for the reporter to fill in
Time, date and location of the incident	Time, date and location of the perceived error
Discovered by	Patient/relative, personnel, other
Patient affected?	Yes/No
Description of the incident	Narrative section
Immediate actions taken	Description of actions taken by the reporter
Contributing factors/situation description	Situation description
Probable causes/cause description	Factors contributing to the incident
Occurrence of harm/consequences	Description of consequences
Severity of outcome/degree of severity	Harm, categorized from none to disastrous (e.g. death of patient, system collapse)
Risk and consequence estimate/risk of recurrence	Risk of recurrence, categorized from low to high
Suggestions for improvements	Narrative section

Data analysis

Data were first inductively analysed using manifest and summative qualitative content analysis.⁷ All incident reports were transcribed and read through several times, and text describing the errors was identified as meaning units. This text was then condensed, without changing its meaning. Condensed meaning units with similar content were placed together and given a name, i.e. sub-categories. Later, sub-categories with similar content were grouped together into categories based on their content (see Table 2). The categorized reports were read independently by two people, and if opinions about categorization varied, consensus was reached after discussion. This analysis yielded five main categories and 13 sub-categories (see Table 2). The data were then analysed using a standard package (Statistical Package for the Social Sciences). The main categories of reported errors were compared for differences between incoming and outgoing incident reports using a chi-square test. Observed and estimated frequencies and adjusted standardized residuals were used. Cell-by-cell comparison of observed and estimated expected frequencies was used to reveal the nature of the dependence.⁸

Table 2

Example of the data analysis process

Original text	Condensed meaning unit	Sub-category	Category
There are no available doctors' appointments at the GP's office for me to give the patient	The GP's office has no doctors' appointments available for the nurse at SHD	Insufficient accessibility – other health-care provider	Insufficient accessibility
SHD does not answer the telephone for about 30–40 minutes, and the patient finally gives up	The patient cannot get in touch with SHD	Insufficient accessibility - SHD	Insufficient accessibility

Results

Errors that lead to an incident report

All sites together received a total of 1,012,988 telephone calls during 2007, a rate of one incident report for every 761 telephone calls. The most common category was accessibility problems (41%), e.g. issues of availability problems at other health-care providers. Incorrect assessment (25%) describes situations such as giving the caller the wrong advice or referring them to the wrong level of care or to a health-care provider in the wrong area. The technical problems category (13%) describes problems and malfunction in the software or telephone equipment, while the routines/guidelines category (15%) describes errors based on failure to follow routines or a lack of routines/guidelines. The information and communication category (6%) describes errors by which someone – the telenurse, patient or health-care staff – perceives that they have been treated in an unpleasant manner or have received unclear information. The most common outgoing incident report concerned accessibility problems: 145 out of 165 described insufficient accessibility at the primary health-care provider, 9 described low accessibility at the specialist and 11 did not specify which other health-care provider the incident report concerned. The most common incoming incident report regarded incorrect assessment. In the incorrect referral sub-category, 36 stated that too-high a level of care was recommended, 20 that too-low a level of care was recommended, and ten did not state this (total 66) (see Table 3).

Table 3

Overview of categories, sub-categories, frequency and incoming or outgoing incident report (Question 1)

Categories and sub-categories	Total (n)	Percent	Incoming	Percent	Outgoing	Percent
Accessibility problems	183	41	18	10	165	61
Insufficient accessibility – other health-care provider	165	37	0	0	165	61
Insufficient accessibility – SHD	18	4	18	10	0	0
Incorrect assessment	114	25	106	58	8	3
Incorrect advice	26	6	26	14	0	0
Incorrect referral – level of care	73	16	66	36	7	3
Incorrect referral – geographic location	15	3	14	8	1	0.4
Technical problems	60	13	4	2	56	21
Software	19	4	0	0	19	7
Telephony	41	9	4	2	37	14
Routines/guidelines	67	15	38	21	29	11
Does not follow formal guidelines	48	11	31	17	17	6
Lack of or inadequate routines	19	4	7	4	12	5
Information and communication	28	6	17	9	11	4
Nurse experienced unpleasant encounter with patient	2	0.4	0	0	2	0.7
SHD nurse complains about encounter with other collaborator	9	2	0	0	9	3
Collaborator complains about encounter with SHD	4	0.9	4	2	0	0
Patient experienced unpleasant encounter with SHD	13	3	13	7	0	0
	452	100	183	100	269	100

Reporter of error

There were 183 incoming incident reports (41% of the total) and 269 (60%) outgoing reports. None of the incident reports described errors within the reporter's own organization. The origin of the incident reports are shown in Table 4.

Table 4

Origin of the incident reports (Question 2)

Incident report origin	Frequency (%)	Most frequently reported (n)
Outgoing	269 (60)	Accessibility problem (165)
Incoming	183 (41)
– Patients	37 (8)	Accessibility (16)
– District health centre	43 (10)	Incorrect assessment (24)
– Specialist clinic	45 (10)	Incorrect assessment (30)
– Emergency room	21 (5)	Incorrect assessment (19)
– On-call doctor centre	31 (7)	Incorrect assessment (21)
– Municipality	3 (0.7)	Accessibility (2)
– Emergency dispatch	3 (0.7)	Incorrect assessment (2)
Total	452 (100)

Differences between incoming and outgoing incident reports

There was a significant difference between incoming and outgoing reports regarding the main category (P < 0.001). Table 5 shows that there were more outgoing incident reports concerning accessibility problems and technical problems than independence would predict, and that there were more incoming incident reports concerning incorrect assessment.

Table 5

Chi-squared test regarding incoming/outgoing incident reports

	Routines/guidelines	Incorrect assessment	Accessibility problems	Technical problems	Information and communication
Outgoing
Observed frequency	29	8	165	56	11
Expected frequency¹	40	68	109	36	17
Standardized adjusted residuals²	−2.9	−13.2	11.0	5.7	−2.3
Incoming
Observed frequency	38	106	18	4	17
Expected frequency¹	27	46	74	24	11
Standardized adjusted residuals²	2.9	13.2	−10.9	−5.7	2.3
Total	67	114	183	60	28

¹Estimated expected frequencies for testing independence;

²A standardized adjusted residual that exceeds about 2 or 3 in absolute value indicates lack of fit of the null hypothesis in that cell¹⁸

Discussion

The most common reported error concerned accessibility problems (41%). This could be explained by SHD's function, i.e. to refer callers to other care providers to manage patient flows, and these could be classified as process errors (cf. Dovey et al. ⁹). In the present study, 6% of all errors dealt with information and communication errors. This is similar to the results of other studies regarding primary health care,^10–12 while another study¹³ showed a substantially higher proportion of communication errors (71%). Approximately 13% of the incident reports in our study concerned technical errors, i.e. problems in the software or telephone equipment, while a previous study on primary health care in the USA reported technical errors at a rate of 24%. However, the USA study classified errors such as misdiagnosis and adverse drug events as technical errors¹⁴ whereas misdiagnosis was classified as incorrect assessment in our study, which could explain the higher number.

Our results showed that the most commonly reported error from SHD to other health-care providers concerned accessibility problems, i.e. the telenurses reported limited possibilities for referring the caller to other care providers. The most commonly reported error from other health-care providers to SHD concerned incorrect assessment, and the most commonly reported error from patients concerned accessibility problems. Patients in our study only contributed 8% of the incident reports, which may be due to the fact that patients in Sweden conventionally make their complaints to the Patient Safety Committee, which would mean that these were not included in our study. A previous study¹⁴ of patient-reported errors in primary care showed that 29% of the errors dealt with complaints about poor access to clinicians. The aim of the SHD is to increase accessibility to health care for the population.¹⁵ However, as shown in our study, telenurses have limited possibilities for referring the callers to the appropriate level of care. In other words, politicians have given telenurses an aim but have not given them the means of fulfilling it. This decreased possibility to refer the caller to the appropriate caregiver may place strain on telenurses and contribute to their feeling of not being able to perform their work tasks as expected.

One interpretation of the results concerning incoming and outgoing incident reports is that telenurses sometimes experience accessibility problems at the appropriate level of care, and thus over-triage callers to a higher level of care or under-triage them to a lower level of care. The fact that telenurses experience problems in their work because they do not have anyone to refer the caller to has been described previously.^4,5,16 A study by Wahlberg et al.⁴ showed that a lack of health-care resources sometimes makes the telenurse over-triage callers, for instance advising those in need of a primary health-care physician to go to a hospital emergency department. Another reason for over-triage may be that telenurses are afraid of being reported if the caller does not receive care within an appropriate time frame. This over/under-triage, in turn, may cause other care providers to report incorrect assessment, e.g. incorrect referral to SHD. Incorrect assessment and over/under-triage may also be due to the protocols the CDSS is based on.¹⁷ Richards et al. ¹⁸ suggested that the CDSS is detrimental to patient outcomes and in their study the CDSS had a negative outcome on the number of after-hours appointments, the number of accident and emergency visits and the number of return consultations. A CDSS has been shown to extend the triage process, possibly because of the lengthy algorithms involved or the fact that most telenurses are unfamiliar with practice nursing.¹⁹ According to Randell et al.,¹⁷ CDSS have been introduced in health care without adequate evaluation and whether they really contribute to increased patient safety remains unanswered.¹⁷

Analysis of incident reports may be considered a form of system analysis, and may contribute to a safer health-care system.²⁰ Surprisingly, we did not find any incident reports regarding the reporter's own organization, i.e. there were only reports of errors made by others. Could this be explained by a culture of shame and blame? All incident reports had a direction and reporting errors made by others might be seen as a sign of frustration. Our view is that this reporting of perceived errors, i.e. those made by others, could contribute to the increase in territorial thinking described earlier.²¹ The implementation of new partners such as centralized telenursing may affect established health-care providers positively or negatively, which may lead to territorial thinking, perhaps revealed in the complaints of others.²¹

Methodological considerations

Analysing incident reporting is merely ‘a window to the system’²¹ and not a search for the cause, but it does provide an opportunity to identify and describe errors in the organization. Earlier studies^22,23 on incident reporting have shown that health-care personnel are reluctant to report errors, and there is therefore a risk that our study describes only some of the errors. We have only described these perceived errors and have not followed up the incident reports. The strength of the present study is that the sample was quite large, and during the analysis, two of the authors read and discussed all the categorized errors until agreement had been reached.

Conclusions

Telenurses at SHD perceive problems in referring the caller to their local health-care centre or specialist clinic due to poor accessibility, and thus tend to over- or under-triage the patient. The SHD sends incident reports to the local health-care centre or specialist clinic concerning low accessibility, and other care providers send incident reports to the SHD concerning incorrect assessment. This information loop could increase territorial thinking and inhibit cooperation between the SHD and other health-care providers. The implications for practice are that poor accessibility is a matter that should be addressed and that the reasons for incorrect assessment need to be explored.

Footnotes

Acknowledgements

We are grateful to the Director of Operation at the SHD for the incident reports, and to Hans Högberg for statistical help. Grants were received from the Swedish Research Council (grant no 522-2005-7461) and the Faculty of Medicine, Uppsala University.

References

Marklund

, Ström

, Månsson

, Borgquist

, Baigi

, Fridlund

Computer-supported telephone nurse triage: an evaluation of medical quality and costs. J Nurs Manag 2007;15:180–7

Ernesäter

, Holmström

, Engström

Telenurses' experiences of working with computerized decision support: supporting, inhibiting and quality improving. J Adv Nurs 2009;65:1074–83

Holmström

Decision aid software programs in telenursing: not used as intended? Experiences of Swedish telenurses. Nurs Health Sci 2007;9:23–8

Wahlberg

, Cedersund

, Wredling

Telephone nurses' experience of problems with telephone advice in Sweden. J Clin Nurs 2003;12:37–45

Holmström

, Dall'Alba

‘Carer and gatekeeper’ - conflicting demands in nurses' experiences of telephone advisory services. Scand J Caring Sci 2002;16:142–8

Lattimer

, George

, Thompson

, Safety and effectiveness of nurse telephone consultation in out of hours primary care: randomised controlled trial. The South Wiltshire Out of Hours Project (SWOOP) Group. BMJ 1998;317:1054–9

Hsieh

, Shannon

SE.

Three approaches to qualitative content analysis. Qual Health Res 2005;15:1277–88

Agresti

Categorical Data Analysis. 2nd edn. New York: Wiley-Interscience, 2002

Dovey

, Meyers

, Phillips

, A preliminary taxonomy of medical errors in family practice. Qual Saf Health Care 2002;11:233–8

10.

Hoffmann

, Beyer

, Rohe

, Gensichen

, Gerlach

FM.

‘Every error counts’: a web-based incident reporting and learning system for general practice. Qual Saf Health Care 2008;17:307–12

11.

Kostopoulou

, Delaney

, Munro

CW.

Diagnostic difficulty and error in primary care – a systematic review. Fam Pract 2008;25:400–13

12.

Rosser

, Dovey

, Bordman

, White

, Crighton

, Drummond

Medical errors in primary care: results of an international study of family practice. Can Fam Physician 2005;51:386–7

13.

Fernald

, Pace

, Harris

, West

, Main

, Westfall

JM.

Event reporting to a primary care patient safety reporting system: a report from the ASIPS collaborative. Ann Fam Med 2004;2:327–32

14.

Kuzel

, Woolf

, Gilchrist

, Patient reports of preventable problems and harms in primary health care. Ann Fam Med 2004;2:333–40

15.

Swedin

Vårdråd direkt: sjukvårdsrådgivningar i samverkan: slutrapport från utredningen om nationellt samordnad sjukvårdsrådgivning. [Swedish Health Care Direct: health call centres in cooperation: final report from the investigation of national co–ordinated health care.] Stockholm: Landstingsförbundet, 2003 [Swedish]

16.

Holmström

, Höglund

AT.

The faceless encounter: ethical dilemmas in telephone nursing. J Clin Nurs 2007;16:1865–71

17.

Randell

, Mitchell

, Dowding

, Cullum

, Thompson

Effects of computerized decision support systems on nursing performance and patient outcomes: a systematic review. J Health Serv Res Policy 2007;12:242–9

18.

Richards

, Meakins

, Tawfik

, Nurse telephone triage for same day appointments in general practice: multiple interrupted time series trial of effect on workload and costs. BMJ 2002;325:1214

19.

Richards

, Godfrey

, Tawfik

, NHS Direct versus general practice based triage for same day appointments in primary care: cluster randomised controlled trial. BMJ 2004;329:774

20.

Vincent

CA.

Analysis of clinical incidents: a window on the system not a search for root causes. Qual Saf Health Care 2004;13:242–3

21.

Hultberg

, Lönnroth

, Allebeck

Evaluation of the effect of co-financing on collaboration between health care, social services and social insurance in Sweden. Int J Integr Care 2002;2:e09

22.

Evans

, Berry

, Smith

, Attitudes and barriers to incident reporting: a collaborative hospital study. Qual Saf Health Care 2006;15:39–43

23.

Currie

, Watterson

Challenges in delivering safe patient care: a commentary on a quality improvement initiative. J Nurs Manag 2007;15:162–8