Feasibility of the SINERGIAPS (“Sinergias entre profesionales y pacientes para una Atención Primaria Segura”) intervention for improving patient safety in primary care

Abstract

Aim

The primary aim was to examine the feasibility of intervention delivery and of trial procedures. Secondary aims were to study the intervention uptake; its acceptability and perceived utility; and its potential to improve safety culture and avoidable hospital admissions.

Methods

We conducted a 3-month, single-arm feasibility study in 10 primary care (PC) centres in Spain. Centres received information regarding patients’ experiences of safety (through the Patient Reported Experiences and Outcomes of Safety in Primary Care [PREOS-PC] questionnaire), and were instructed to plan safety improvements based on that feedback. We used a bespoke online tool to recruit PC professionals, collect patient feedback, and deliver it to the centres, and to collect outcome data (patient safety culture [Medical Office Survey on Patient Safety Culture, MOSPSC questionnaire]). We measured recruitment and follow-up rates and intervention uptake (based on the number of safety improvement plans registered). We conducted semistructured interviews with 9 professionals to explore the intervention acceptability and perceived utility.

Results

Of 256 professionals invited, 120 (47%) agreed to participate, and 97 completed baseline and postintervention measures. Of 780 patients invited, 585 (75%) completed the PREOS-PC questionnaire. Five of 10 centres (50%) designed an improvement plan, providing 27 plans in total (range per centre, 1–14). The intervention was perceived as a novel strategy for improving safety, although the healthcare professionals identified several factors limiting its acceptability and utility: lack of feedback at the individual professional level; potentially unrepresentative sample of patients providing feedback; and number of educational materials deemed overwhelming.

Discussion

It is feasible to deliver the proposed intervention so long as the identified limitations are addressed.

doctor–patient relationship, health services research, medical errors/patient safety, mixed methods, practice management, primary care

Key Messages

• We piloted a patient feedback intervention to improve patient safety in primary care.

• Recruiting and retaining professionals were challenging and resource intensive.

• It was feasible to deliver the intervention to all the 10 centres.

• Uptake was moderate (only 5 out of 10 centres designed improvement actions).

• Scaling the intervention is feasible after introducing major modifications in its design.

Background

In primary care (PC), 2%–3% of all consultations result in a patient safety incident, with 1 in 25 causing serious harm.¹ Such incidents are frequently related to diagnoses (either delayed or missed)² or to treatment (delayed or inappropriate).³ In Spain, the country with the highest PC consultation rates in Europe,⁴ around 3 million adverse events occur in the PC setting each year,⁴ costing up to 1 billion euros yearly.⁵ Almost two-thirds of these incidents (64%) are preventable.⁴

As highlighted by the World Health Organization, increasing patient involvement in patient safety is key to reducing adverse events.^6,7 Patients are the common element across the various settings, organizations, and professionals generally involved in their healthcare, and therefore, they are ideally suited to reflect on the healthcare they receive.^8,9 Available evidence from hospitals affirms the effectiveness of patient feedback interventions.^10–12 In the PC setting, despite the recent proliferation of studies,^13,14 evidence of effectiveness on safety outcomes from fully powered, phase III trials is still lacking. To contribute to addressing this knowledge gap, we developed the SINERGIAPS (“Sinergias entre profesionales y pacientes para una Atención Primaria Segura”/Synergizing providers and patients for Safer Primary Care) intervention, a low-cost and scalable theory-based, online intervention for improving patient safety in PC centres. It is based on the use of patient feedback collected through the Spanish version of the Patient Reported Experiences and Outcomes of Safety in Primary Care (PREOS-PC) questionnaire,¹⁵ a tool initially developed¹⁶ and extensively researched in the United Kingdom.^17–22

The primary aims of this mixed-method feasibility study were to examine the feasibility of intervention delivery (concerning both the administration of the PREOS-PC questionnaire, and the feedback to practices); and of trial procedures (recruitment methods and outcome data collection). Secondary aims included to study the intervention uptake; to explore the intervention acceptability and perceived utility by recipients (PC providers); and to examine the potential impact of the intervention on improving patient safety culture and reducing avoidable hospitalizations.

Methods

Study design

From May to August 2019, we piloted the SINERGIAPS intervention in a 3-month, single-arm (pre–post) feasibility trial in 10 PC centres in Majorca, Spain.

Recruitment of PC centres and staff

We organized face-to-face meetings in 10 PC centres purposefully selected to ensure variation in terms of number of registered patients, rurality, and patients’ economic status based on data supplied by the Balearic Islands’ Regional Health Service (see Supplementary Material 1). The meetings aimed to present the study to all of the PC professionals in the centre (via a 30-min presentation followed by a question-and-answer session) and to obtain written informed consent to enrol the centres.

Study participants

We sent email invitations to all PC professionals working in the centres (doctors, nurses, nurse assistants, and administrative staff). We excluded professionals on sick leave and those not planning to be working at their centre during the study period.

Sample size

Our sample size calculation aimed at estimating the precision of the follow-up rate (proportion of professionals completing the MOSPSC questionnaire at postintervention out of the total number of professionals enrolled)—the primary outcome of this feasibility study (see “Outcomes Measures”). Assuming an average of 26 PC professionals per centre,²³ we estimated that recruiting 10 centres would result in approximately 260 professionals taking part in the study (assuming that all professionals in the centres would participate). With 260 participants, this feasibility study is powered to detect 80% follow-up within 95% confidence intervals of 75.1%–84.9%.

Description of the SINERGIAPS intervention

The SINERGIAPS intervention has been described in detail elsewhere.²⁴ In brief, it is based on the Clinical Performance Feedback Intervention Theory, which states that behaviour is regulated through comparisons with standards or goals, and that feedback can draw attention to existing gaps.²⁵ The intervention was codesigned with 43 PC professionals who participated in a qualitative study that involved 4 focus groups and 3 semistructured individual interviews.²⁶ The intervention, delivered fully through a bespoke online tool, involved 3 stages (Fig. 1):

Fig. 1.

Description of the SINERGIAPS theory-based online intervention centred around the systematic patient feedback for improving patient safety in PC, piloted in 10 PC centres in Spain (2019).

• a)

  Measurement: All patients in waiting rooms were consecutively invited to complete the Spanish online version of the PREOS-PC Compact questionnaire¹⁵ by members of our research team (MJS-R and IR-C) using tablet computers. We administered 50 questionnaires per practice (based on the number of responses needed to achieve 0.7 reliability in scale scores¹⁶). All patients received plain language Patient Information Sheets and consent forms. From all the patients invited, we registered their decision to participate, and age and gender.

• b)

  Feedback: From 5 May to 5 August 2019, all participating centres were given access to a password protected, centre-specific feedback report, accessed through a secured link to our online platform. The report (Supplementary Material 2), similar to reports from previous patient feedback interventions,^27–30 had been previously codesigned with PC professionals who took part in 4 workshops.²⁶ The report was automatically generated using a bespoke online tool, provided real-time information about the performance of each PC centre in comparison with the other participating centres. A set of educational materials and online resources (repository of publicly available written materials, infographics, and videos from local, national, and international healthcare organizations) were also provided through the online platform. Follow-up calls were made by a member of our team (MJS-R) to ensure all centres had access to the intervention materials and to resolve technical problems, when needed.

• c)

  Action planning and implementation: PC centres were instructed to create an Action Planning Team (APT) responsible for receiving the Feedback Report, considering the area(s) that should be addressed, and designing an action plan for safety improvement. The APTs were prompted to register their plans in a structured form available via the online tool. The intervention was meant to be self-contained, and therefore the researchers did not support centres in the analysis of the feedback report nor in the action planning and implementation.

Data collection

The Spanish version of the Medical Office Survey on Patient Safety Culture (MOSPSC) questionnaire (a valid and reliable tool for assessing patient safety culture in Spain³¹) was administered online during the preintervention period and after the 3-month intervention period to all participating PC professionals. Safety culture questionnaires are frequently used for evaluating patient safety interventions and programs, and for tracking changes over time.³²

Baseline characteristics of the participating PC centres and the number of avoidable hospital admissions per centre were extracted from electronic health records by an information technology specialist at the Balearic Islands’ Regional Health Service (see Acknowledgements). The definition for avoidable hospital admissions used in this study stems from the Prevention Quality Indicators proposed by the Agency for Healthcare Research and Quality.³³ Avoidable admissions for asthma, chronic obstructive pulmonary disease, congestive heart failure, angina, complications of diabetes, and dehydration in those aged 65 and over were identified using a set of prespecified ICD-9 codes (available in Supplementary Material 3) previously used elsewhere.³⁴ For each participating centre, we extracted the total number of patients registered, the number of avoidable hospital admissions registered during the 3 months prior the start of the intervention (baseline measure), and the number of avoidable hospital admissions registered during the 3 months after the intervention (postintervention measure).

Outcome measures

The primary outcome was professionals’ follow-up rate, which was measured as the proportion of PC professionals who completed the MOSPSC questionnaire at the baseline (denominator) and postintervention (numerator) times. Secondary outcomes related to trial feasibility included the recruitment rate (number of PC professionals accepting to participate/total number of PC professionals invited) and patients’ rate of response to the PREOS-PC questionnaire (number of patients completing the questionnaire/total number of patients invited). Secondary outcomes related to the impact of the intervention included changes in patient safety culture, and in avoidable hospital admission rates.

Statistical analyses

We conducted descriptive statistical analyses, calculating frequencies and percentages to estimate recruitment and follow-up rates. We calculated the median and interquartile range (IQR) to describe the characteristics of the centres. We used paired t-tests to examine pre–post differences in mean scores (overall and domain-specific) of the MOSPSC questionnaire at the PC professional level. Nonparametric tests (Wilcoxon matched-pairs signed-rank test and Wilcoxon rank-sum test) were used to examine pre–post differences in avoidable hospitalization rates at the centre level, and to explore differences between centre characteristics (size and patient-reported patient safety) according to intervention uptake. All analyses were conducted with STATA v12.

Embedded qualitative study

We conducted qualitative, post-trial, semistructured interviews with 9 PC professionals to evaluate the acceptability and perceived utility of the intervention, and to identify implementation barriers. Professionals were purposefully selected to ensure variation in terms of professional roles. We used an interview guide that was flexible in order to elicit perceptions, suggestions, and opinions related to the proposed intervention (Supplementary Material 4). The interviews were carried out by telephone during July and August of 2019. On average they lasted 24 min (range: 13–48 min). They were recorded, transcribed verbatim, and analysed using content analysis.³⁵ Two researchers (MJS-R and GP-M) conducted the initial analysis, coding inductively in an iterative process. All research team members discussed the data and categories until consensus was reached. Along with the description of the themes we present selected quotes (translated into English by a professional translator) to enrich the interpretation of our results.

Results

Recruitment and follow-up rates of PC centres and professionals

All the 10 PC centres initially invited agreed to participate in the study. Of the 256 PC workers invited, 120 (47%) accepted and completed baseline measures (Supplementary Material 5). Of the 120 recruited professionals who completed the baseline measures, around 3 quarters (78%) were women, 40% were doctors, and 26% were nurses (Table 1). The large majority (115/136; 85%) of the professionals invited who did not participate in our study did not respond to our email invitations. The most common reason for not participating among those who did respond was lack of time (15 professionals).

Table 1.

Characteristics of the participating professionals who completed baseline and follow-up measures of the feasibility study (data collection period: May–August 2019).

	Healthcare professionals recruited (n = 120)	Healthcare professionals with completed follow-up measures (n = 96)
	N (%)	N (%)
Age (years)
Mean (SD)	49.06 (9.45)	50.63 (8.82)
18 to <30	3 (2.50)	1 (1.04)
30 to <40	15 (12.5)	9 (9.38)
40 to <55	63 (52.5)	47 (48.96)
55 to 69	39 (32.51)	39 (40.63)
Sex
Men	27 (22.5)	23 (23.96)
Women	93 (77.5)	73 (76.04)
Role
Doctor	48 (40)	41 (42.71)
Nurse	31 (25.83)	21 (21.88)
Administrative	17 (14.17)	15 (15.63)
Nursing assistant	2 (1.67)	1 (1.04)
Doctor manager	8 (6.66)	8 (8.33)
Nurse manager	5 (4.17)	5 (5.21)
Admin manager	7 (5.83)	5 (5.21)
Years at work in the PC centre
0 to <3	23 (10.83)	10 (10.42)
3 to <6	14 (11.67)	14 (14.58)
6 to <11	32 (26.67)	27 (28.13)
11 to <20	34 (28.33)	29 (30.21)
>20	17 (14.17)	16 (16.67)
List size
≤500	11 (9.17)	10 (10.42)
501–1,000	0 (0)	2 (2.08)
1,001–1,500	18 (15)	10 (10.42)
1,501–2,000	60 (50)	48 (50)
>2,000	31 (25.83)	26 (27.08)

Postintervention follow-up questionnaires were completed by 96 of the 120 PC professionals (81%). No significant differences in terms of age, sex, role, years at work in the centre, and list size were observed between professionals who completed the postintervention measures vs those who did not. Reaching our target follow-up rate of 80% required sending up to 5 reminders and combining electronic with paper-based questionnaires (a research assistant visited the participating centres and offered the professionals to complete a paper-based version, subsequently uploading the results into our platform). This resulted in having to extend the data collection period by 2 months.

Patients completing the PREOS-PC questionnaire

Of the 780 patients invited, 585 (75%) agreed to complete the questionnaire. Of those, 493 (84%) patients responded to all of its items. Participating patients were similar to the rest of the patients visiting each centre. Around two-thirds (63%) were female, and 24% were aged 65 and over. The questionnaire completion time averaged 15 min.

Feasibility of intervention delivery and intervention uptake

Five centres experienced technical issues and required assistance accessing the intervention materials. After resolving these issues, all of the centres successfully accessed the intervention materials.

Table 2 summarizes the centres’ intervention uptake. All of the centres reported that they created an APT to review the feedback report. On average, APT comprised 3 professionals (the most frequent combinations being 2 doctors plus 1 nurse; or 1 doctor plus 1 nurse and 1 administrative). Only 6 reported that the team successfully met during the study period, and just 5 registered action plans. No differences were observed in PREOS-PC scale scores between the centres that registered action plans and those that did not. However, the median (IQR) number of registered patients was significantly larger (P = 0.04) in centres that registered at least 1 action plan (19,890 [19,629–25,194]) compared with those not registering any plan (6,449 [5,106–9,162]). Similarly, the median (IQR) number of professionals was marginally larger (Wilcoxon rank-sum test P = 0.06) in centres that successfully registered at least 1 action plan (35 [27–41]) than in centres not registering any plan (11 [9–19]). No differences were observed in terms of the number of professionals recruited or the recruitment rate. No relationship was identified between experiencing teething technological issues accessing the intervention and the registration of action plans.

Table 2.

Intervention uptake and summary of the patient-reported patient safety report fed back to the 10 PC centres participating in the feasibility study (data collection period: May–August 2019).^g

Centre	Intervention uptake				Patient-reported patient safety (PREOS-PC scores∗)
	Successfully accessed to the intervention materialsh	Action plan team formedi	Action plan team metj	Number of action plans registeredk	Overall ratea	Practice activationb	Patient activationc	Patient experiencesd	Harm (severity)e	Harm (burden)f
Centre 9	Yes	Yes	Yes	4	89.80 (11.86)	85.42 (18.53)	37.50 (35.96)	94.95 (11.34)	95.38 (10.92)	97.33 (8.66)
Centre 6	Yes	Yes	Yes	3	89.00 (12.66)	82.25 (21.07)	52.19 (40.72)	92.44 (9.19)	96.13 (9.52)	97.67 (7.90)
Centre 3	Yes	Yes	No	0	86.89 (14.43)	80.19 (18.73)	34.48 (34.01)	92.72 (11.79)	97.36 (8.48)	98.15 (5.59)
Centre 2	Yes	Yes	Yes	0	86.60 (12.87)	84.42 (20.85)	46.47 (37.60)	93.88 (10.36)	97.25 (9.96)	98.67 (5.42)
Centre 7	Yes	Yes	Yes	1	85.80 (16.91)	77.04 (19.80)	49.68 (38.20)	91.44 (13.54)	95.00 (12.63)	95.33 (10.93)
Centre 8	Yes	Yes	Yes	14	84.12 (16.75)	78.92 (19.35)	39.24 (36.25)	93.51 (10.53)	95.96 (10.88)	98.86 (4.42)
Centre 10	Yes	Yes	No	0	83.60 (14.39)	72.88 (16.92)	36.05 (38.50)	90.26 (13.52)	92.50 (17.81)	92.83 (21.16)
Centre 1	Yes	Yes	No	0	83.00 (15.94)	84.00 (17.46)	44.60 (37.68)	91.04 (11.42)	93.88 (16.96)	97.00 (9.18)
Centre 4	Yes	Yes	No	0	81.80 (18.48)	78.71 (20.96)	48.48 (37.21)	91.72 (13.82)	94.13 (16.77)	94.17 (10.55)
Centre 5	Yes	Yes	Yes	6	78.91 (21.52)	74.28 (21.80)	38.95 (33.54)	86.85 (17.41)	92.39 (14.90)	94.93 (12.85)

CI, confidence interval.

Intracluster correlation = 0.03 (95% CI = 0–0.07).

Intracluster correlation = 0.04 (95% CI = 0–0.13).

Intracluster correlation = 0.01 (95% CI = 0–0.06).

Intracluster correlation = 0.00 (95% CI = 0–0.04).

Intracluster correlation = 0.00 (95% CI = 0–0.03).

Intracluster correlation = 0.02 (95% CI = 0–0.08).

Data are mean (SD).

The participating centres could successfully access to the intervention materials (feedback report, educational materials and action planning template) through the online platform (yes/no).

The participating centres could successfully assemble an Action Plan Team (yes/no).

The Action Plan Teams of the participating centres met at least once to review the feedback report and discuss potential safety improvement actions (yes/no).

Number of safety improvement actions registered by the Action Plan Teams in the online platform template.

Supplementary Material 6 lists the actions proposed by the participating centres, most of which address treatment-related safety problems. Organizing clinical sessions for providing further training to PC providers was the most frequently proposed type of action. Workshops with patients and carers to improve polypharmacy management and to reduce patient requests for unnecessary diagnostic tests were also proposed. Other actions included setting up suggestion boxes and reorganizing agendas to shorten waiting lists.

Impact of the intervention on safety culture and avoidable hospitalizations

Compared with the baseline, postintervention MOSPSC scores (Table 3) were significantly higher for the overall safety culture score (3.36/5 at baseline vs 3.44/5 at postintervention (a 2% increase); P = 0.01) and for the domain “Office processes and standardization for non-healthcare staff” (0.37/1 vs 0.45/1 [22% increase]; P = 0.02), indicating an increased level of safety culture. No significant improvements were observed for the remaining domains.

Table 3.

Impact of the intervention on safety culture measured with the Medical Office Survey on Patient Safety Culture (MOSPSC) domain-specific and overall scores during the feasibility study (data collection period: May–August 2019).

	Preintervention		Postintervention		Post–pre intervention difference
	Mean	95% CI	Mean	95% CI	Mean	95% CI
Patient safety and quality issues (n = 95)	0.5525	0.5044; 0.6007	0.5838	0.5348; 0.6327	0.0312	−0.0203; 0.0827
Information exchange with other settings (n = 87)	0.3142	0.2325; 0.3959	0.3180	0.2411; 0.3949	0.0038	−0.0808; 0.0885
Teamwork (n = 96)	0.7049	0.6367; 0.7730	0.7240	0.6591; 0.7888	0.0191	−0.0353; 0.0735
Work pressure and pace (n = 96)	0.1354	0.0973; 0.1735	0.1250	0.0854; 0.1646	−0.0104	−0.0499; 0.0291
Nonhealthcare staff training (n = 87)	0.4885	0.4051; 0.5719	0.5000	0.4171; 0.5829	0.0115	−0.0775; 0.1005
Healthcare staff training (n = 84)	0.5754	0.4934; 0.6574	0.6071	0.5237; 0.6905	0.0317	−0.0481; 0.1116
Office processes and standardization for nonhealthcare staff (n = 96)∗	0.3654	0.2980; 0.4329	0.4462	0.3743; 0.5180	0.0807	0.0109; 0.1505
Office processes and standardization for healthcare staff (n = 96)	0.4410	0.3748; 0.5071	0.4740	0.4031; 0.5448	0.0330	−0.0374; 0.1033
Communication openness (n = 96)	0.5434	0.4709; 0.6159	0.5972	0.5233; 0.6711	0.0538	−0.0285; 0.1362
Patient care tracking/follow-up (n = 92)	0.7274	0.6605; 0.7942	0.6966	0.6342; 0.7589	−0.0308	−0.0898; 0.0283
Communication about error. Nonhealthcare staff (n = 96)	0.5720	0.5023; 0.6418	0.5894	0.5146; 0.6642	0.0174	−0.0516; 0.0863
Communication about the error healthcare staff (n = 96)	0.6163	0.5485; 0.6841	0.6415	0.5667; 0.7163	0.0252	−0.0414; 0.0918
Leadership support for patient safety (n = 71)	0.5282	0.4427; 0.6136	0.5528	0.4596; 0.6461	0.0246	−0.0524; 0.1016
Organizational learning (n = 95)	0.6789	0.5986; 0.7593	0.6982	0.6197; 0.7768	0.0193	−0.0536; 0.0922
Overall perceptions of patient safety and quality (n = 95)	0.5474	0.4681; 0.6266	0.5956	0.5216; 0.6696	0.0482	−0.0143; 0.1108
Overall ratings on quality (n = 96)	0.7729	0.7221; 0.8237	0.8125	0.7606; 0.8644	0.0396	−0.0139; 0.0930
Overall rating on patient safety (n = 96)	0.7917	0.7089; 0.8744	0.8229	0.7452; 0.9007	0.0313	−0.0490; 0.1115
Overall safety culture score (n = 96)∗	3.3595	3.2568; 3.4422	3.4363	3.3484; 3.5242	0.0868	0.0268; 0.1467

CI, confidence interval.

∗Statistically significant (P < 0.05) post–pre intervention differences observed (Wilcoxon matched-pairs signed-rank test).

The improvements in the overall safety culture score were significantly higher (Wilcoxon rank-sum test P = 0.047) in centres that registered at least 1 improvement plan (median [IQR] pre–post difference = 0.11 [0.09 to 0.16]) than in those not registering any plan (0.03 [−0.06 to 0.05]).

The median (IQR) rate of avoidable hospitalizations per 1,000 registered patients was lower during the 3 months after the intervention (0.45 [0.33–0.83]) than during the 3 months before the onset of the intervention (0.78 [0.70–0.9]), although the difference was not statistically significant (P = 0.07) (Supplementary Material 7). No differences in rates of avoidable hospitalizations were observed between centres that registered safety improvement plans and those that did not.

Results from the qualitative interviews

Nine PC professionals from 7 centres participated in the semistructured interviews (5 doctors, 2 nurses, 1 paediatrician, and 1 administrative staff member, and 7 of whom were women). Two main categories were identified:

• i.

  Perceived utility and acceptability of the intervention. Participants perceived the SINERGIAPS intervention as a novel strategy that could produce long-term safety improvements by raising their awareness about patient safety.

Being able to meet with the team to comment and assess those things that had gone “well” and “wrong”, I think it is very good for all of us, I mean, for the professionals but also for our patients. (49-year-old female administrator)

The method for collecting patient feedback was perceived as acceptable: professionals did not identify disruptions to their activities as a result of the researchers administering patient questionnaires in waiting rooms. Though they found patient feedback reports useful (as they provided rich, relevant, and actionable information for improving safety in PC centres), they suggested that reports could be improved by recording performance at the individual professional level (in addition to at the centre level). They suggested including additional information about the methods used to collect and analyse patient feedback, and they raised concerns about the accuracy and representativeness of the feedback data.

I had the feeling that the sample was too small to actually be valuable. I think this is not helping us to improve our work. (62-year-old female nurse)

The educational materials accompanying the feedback report were perceived as overly thorough and comprehensive, and it was suggested that more succinct materials be produced.

Professionals reported that action planning meetings had been limited by the fact that the feasibility study took place during the summer.

Another problem is the time of year (…) it caught us in the middle of August, when almost everyone was on leave. (44-year-old female doctor)

• ii.

  Strategies to increase the participation rate of professionals in future trials: participants attributed the low participation rate to heavy workloads and a low level of a teamwork culture. For future trials, they suggested involving a professional from each centre as a leader to support recruitment and data collection, and to raise awareness about patient safety.

Conclusions

In this study, we observed that it was feasible to implement SINERGIAPS—a novel, online, theory-based patient feedback intervention—to improve patient safety in PC centres. Although several adjustments to the intervention materials are needed to maximize its uptake, SINERGIAPS was perceived as acceptable and useful by PC professionals and appears to be feasible for scaling up to a phase III trial.

Strengths and limitations

We recruited a large number of participants and used a mixed-methods approach to examine the feasibility, acceptability, perceived utility, and potential impact of the SINERGIAPS intervention, as well as related implementation barriers. However, our study has some limitations. First, the fact that we measured safety improvements as planned, rather than as executed, limited our ability to obtain more solid evidence about intervention uptake. Second, the high intervention acceptability and perceived utility observed in the qualitative interviews with PC professionals could be partially due to a potential influence of the researcher who interviewed the participants (i.e. a social desirability bias). Finally, the lack of a control group, the low number of participating centres, and the short follow-up period limited our ability to assess the impact of the intervention on safety culture and avoidable hospitalization rates. Also, due to scarcity of resources, we did not administer the PREOS-PC questionnaire at postintervention, and therefore were unable to explore the potential impact of the intervention on patient-reported patient safety. It is however worth noting that this limitation only affects to one of our secondary objectives (study the potential intervention impact); feasibility studies are usually not designed to determine impact on intervention outcomes (impact on outcomes will be formally assessed as part of our 12-month randomized controlled trial [RCT] currently underway).

Discussion of main findings

There is a growing interest in strategies to support patient involvement to improve patient safety.^12,36–38 A recent systematic review of 42 interventions revealed positive effects of the interventions on patient safety.¹² However, most of the research in this area has focussed on the hospital setting, and ours is one of the first studies examining a patient feedback intervention to support patient safety in PC. The recruitment of professionals was challenging, with a recruitment rate substantially lower than initially expected (47% instead of 100%). Although we reached an adequate (>80%) follow-up rate with the MOSPSC questionnaire, it was only achieved after sending multiple reminders and extending the data collection period, which limited our ability to capture the isolated impact of the intervention. After some technological troubleshooting, the intervention was feasible to administer. However, only 5 centres took safety improvement actions. Opportunity (both physical and social) and motivation are key drivers when it comes to changing behaviour,³⁹ and they are likely to have affected the level of uptake of the SINERGIAPS intervention. We observed a higher level of intervention uptake in larger centres, perhaps because they have more resources available and higher levels of organization, which could facilitate intervention engagement. Capacity limitation is indeed a key driver of intervention uptake according to the Clinical Performance Feedback Intervention Theory.²⁵

Our findings are consistent with those from a similar study in Australia, which investigated the feasibility of a patient safety feedback intervention in PC.¹³ The authors observed that staff deemed the intervention acceptable, and that intervention fidelity was high. They identified 3 types of factors mediating with the implementation of the intervention: practice readiness, utilization of problem solving skills, and agency.¹⁴ According to the Patient Feedback Response Framework, making changes based on patient feedback is a complex multitiered process, which is determined by the existence of normative legitimacy (the belief that listening to patients is a worthwhile exercise), structural legitimacy (staff need adequate autonomy, ownership and resource to enact change), and organizational readiness (as frequently changes requires meso- or macrolevel assistance).⁴⁰

Concerning the potential impact of the intervention, our exploratory analysis suggests that the intervention has the potential to improve safety culture. The observation that improvements were larger in centres that fully implemented the SINERGIAPS intervention would further provide evidence of its potential impact. However, given the limitations in the design of our study, this finding needs to be considered with caution.

In conclusion, it is feasible to deliver an online-based intervention centred around patient aimed at improving patient safety culture in Spanish PC centres. The available evidence suggests that the SINERGIAPS intervention has potential to be perceived as acceptable and useful by PC professionals in Spain. However, a number of modifications should be introduced to optimize its potential impact and uptake. Its effectiveness in improving patient safety will be assessed in an RCT involving 59 PC centres.

Acknowledgements

We thank all the patients and PC professionals who participated in this study. We thank Mr Jerónimo Fiol Renaudin for his support in data management related tasks. We also thank Mr Bartolomé Sastre, from the Balearic Health System Information Technology Department, for his support in extracting avoidable hospitalization data from electronic health records.

Funding

This work was supported by a personal award (Miguel Servet Fellowship CP17/00017) to IR-C awarded by the Instituto de Salud Carlos III (Spanish Ministry of Sciences and Innovation) and co-funded by Fondo Europeo de Desarrollo Regional (FEDER). MJS-R and MAF-D were funded by Instituto de Investigación Sanitaria de las Islas Baleares, grant number FOLIUM17/10 (co-funded by ITS-2017 and PO FSE 2014-2020) and FOLIUM19/05 (founded by ITS-2019-003), respectively. The rest of authors were not granted by any grant or award to develop this work. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Ethical approval

The study was approved by the Research Ethics Committee of the Balearic Islands (CEI IB 07/18).

Conflict of interest

IR-C and JMV codeveloped the PREOS-PC questionnaire, which is licensed by Oxford Innovation Ltd. The rest of the authors report no conflict of interest.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Author notes

These authors contributed equally.