Introduction

It is estimated that approximatly 100 000–200 000 Norwegians have undiagnosed type 2 diabetes (T2D) [1]. The disease can develop slowly and the symptoms can be diffuse, therefore people with undiagnosed T2D have a tendency to underestimate their risk of this disease [2]. It can take 4–6 years before the disease is identified [3]. Nevertheless, with early identification and good treatment, it is possible to prevent or to delay long-term complications [4], such as diabetic retinopathy, neuropathy and nephropathy and cardiovascular disease [5].

For early identification of people with T2D and those at risk of T2D, screening can be performed [6]. Although there is some discussion about the utility of screening for T2D [7], a systematic review concluded that this is cost-effective and could be cost-saving [8]. On the other hand, exposing a large segment of the population to screening for T2D may lead to fear—two systematic reviews found that screening could increase the short-term anxiety [9, 10]. However, participants reported no long-term negative psychological impact [9, 10].

Both the World Health Organization [11] and the American Diabetes Association [12] open up for targeted screening for those with a risk of T2D. In Norway it is recommended to identify people with a high risk of developing T2D (case-finding [13]) [13, 14]. The Norwegian National Guidelines for Diabetes [14] recommend to use the validated Finnish Diabetes Risk Score (FINDRISC) as the first step [15]. If the FINDRISC score is high, a HbA1c (hemoglobin A(1c))-measurement should be performed. The HbA1c-test has several advantages compared with other available tests [16, 17] and is the recommended test for diagnosing T2D in Norway [14]. Because non-European ethnic groups have a different risk profile than ethnic Europeans, the Norwegian Diabetes Association recommends the diabetes risk test developed by Diabetes UK [18, 19] for this group.

Community pharmacy has been described as a low threshold service, due to long opening hours, and easily accessible health personnel that can be reached without an appointment [20]. A Swedish study found that over 70% of pharmacy customers think that pharmacies can motivate people to improve their health [21]. Furthermore, in a systematic review it was found that pharmacies are a feasible location to perform screening for T2D [22]. However, this has not been explored in Norway until now, and few studies have explored the use of HbA1c for screening in a pharmacy setting [23–25]. Before implementation of a complex intervention in a large-scale study, it is recommended to perform a feasibility study to optimize the protocol [26]. Thus, the objective with this study was to test the feasibility of using a diabetes risk assessment tool followed by HbA1-measurement in a Norwegian community-pharmacy setting.

Methods

Results

Pharmacy A recruited 152 participants, pharmacy B recruited 36 participants and pharmacy C recruited 31 participants, which gives a total of 219 participants. Seven participants were excluded due to missing data (the risk assessment was not retained at the pharmacy) and one because he already had been diagnosed with diabetes, giving a final number of 211 participants. The background questionnaire, consisting of variables such as highest completed education, why the participant took the test, and where they had heard about the service, in addition to self-rated health, was completed by 198 (94%) of the participants. One hundred and eighty-seven (89%) filled out FINDRISC, while 24 (11%) filled out Diabetes UK. Both tests were available in a Norwegian and an English version; 189 (94%) participants filled in the form in Norwegian, while 13 (6%) used the English version. All participants completed the diabetes risk test. Table 1 shows the distribution of gender, age groups and levels of education by ethnicity among the participants.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. General characteristics of the participants.

https://doi.org/10.1371/journal.pone.0191316.t001

Risk of developing diabetes

Table 2 shows the distribution of risk of developing diabetes among the participants, self-rated health and HbA1c at inclusion. The median score for self-rated general health at inclusion was 70 (range 20–100, n = 183). Seventeen (9%) of the 198 participants who answered the background questionnaire did not complete the question regarding self-rated health. Thirty-two (15%) participants had HbA1c values < 5.7% (normal), twelve (5.7%) had values between 5.7%—6.4% (pre-diabetes), while three (1.4%) had an HbA1c ≥ 6.5% (cut-off for diabetes) [30]. The three HbA1c values ≥ 6.5% were 6.5%, 8.3% and 10.5%, and the corresponding score at the diabetes risk test were 19, 19 (both FINDRISC) and 18 (UK Diabetes).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Results from the diabetes risk tests by ethnicity (FINDRISC [15] and diabetes UK [18]) and the corresponding statistical risk of developing diabetes type 2 within the next 10 years, self-rated health and HbA1c at inclusion.

https://doi.org/10.1371/journal.pone.0191316.t002

Discussion

This study shows that it is feasible to perform T2D case-finding using a diabetes risk assessment tool followed by HbA1-measurement in a Norwegian community-pharmacy setting. To our knowledge this is the first study in Norway where a screening protocol including both diabetes risk assessment test and measurement of HbA1c has been tested. Previous studies performed in other countries using a comparable protocol confirm the conclusion, however different risk assessment tools [24, 31, 32] were used. In addition, only one study used HbA1c-measurement [24], while glucose-measurement was used in the other studies [31, 32].

By using this protocol both people with moderate to very high risk of developing T2D (47 of 211 participants (22%)) and people with undiagnosed T2D (two of 211 (1%)) were identified. Our results also identified some points in the protocol that need optimization before a large-scale study is conducted.

A reason for offering this type of risk assessment in pharmacy, is that a large number of people visit the pharmacies. In Norway the pharmacies had about 50 million customers in 2015 [33]. The participating pharmacies in this study had an average of 300 customers a day in the inclusion period, which implies a total of potential 54 000 people during the study period. Still, many of the same customers visited the pharmacy multiple times during the inclusion period, thus the potential number of participants for this study was lower than the total number of customers. Surprisingly, only 13% of the participants had heard about this service when visiting the pharmacy, while 70% had read about it either in the newspapers/social media and 12% received information from family and friends. This could mean that that most of the participants were not typical pharmacy customers, but people who visited the pharmacy specifically to take the risk test. The fact that a lagre propotion of the participants were quite young, 80 under 45 years, and 17 under 49 years, supports this finding.

The pharmacists were satisfied with the training-program and the follow-up during the study. Furthermore, the results from EQA and internal quality controls show that the quality of the HbA1c-measurements was good. One study found that use of the HbA1c Point of Care (POC)-instrument was a barrier for the pharmacists, showing that training in the use of the DCA-instrument is an important part of the implementation of the service [24]. In our study online learning courses, a course day and follow-up by Noklus laboratory advisors, all contributed to ensure that the pharmacists felt competent to perform the HbA1c POC-testing at the pharmacy. However, the pharmacists did report that the procedure was time-consuming and that at least two pharmacists had to be at work at the same time to be able to perform the screening protocol undisturbed from the other customers and duties at the pharmacy. There were also periods were the project pharmacists were not available to perform the risk assessment.

There was a big difference in the number of participants recruited by the three pharmacies. Surprisingly, pharmacy A, which included the most participants (152), had a lower average number of customers per day (330), compared to pharmacies B and C, which included 36 and 31 participants, and had 457 and 359 customers per day. An explanation could be that the pharmacy with the lowest number of customers had more time to spend on this project and on recruitment. Moreover, the difference in recruitment could have been due to different levels of enthusiasm among the pharmacists. However an other possible reason could be that a regional newspaper wrote a article about pharmacy A offering the service and that this article was widely spread on social media (Facebook). The two other pharmacies did not receive similar media coverage.

We found no difference in the median score for self-rated health at inclusion and at follow-up, and 82% of the participants were not more concerned about getting diabetes after taking the risk test. Previous studies have also found that attending a screening programme does not have any long-lasting adverse psychological impact [34] and also that screening for T2D has limited impact on self-rated health [10, 35]. However, one study found that participants who screened positive had a poorer general health [34].

We used two different risk assessment tests, following national recommendations. In FINDRISC, it is recommended to measure HbA1c when the risk score is equal or over 15 and statistically one of three develops T2D within the next ten years [14]. While in the diabetes risk test from the Diabetes UK it is recommended to measure HbA1c when the score is equal or over 16 and statistically one of seven risk develops T2D within the next ten years [18]. We cannot see any reason why one chooses such different cut-off values, given that the total test-scores take into consideration the higher risk among non-Europeans. But we have chosen to follow the instructions of each test.

The use of validated measures for diabetes risk and self-rated health [36–38] was a strength of this study. Furthermore, the cut-off limits recommended for the diabetes risk tests was used, and the Norwegian National Guidelines for Diabetes was followed [14, 39, 40]. Additionally, the pharmacies were enrolled in Noklus’ EQA program for HbA1c ensuring that the measurement of HbA1c was quality assured. In Norway, the use of HbA1c POC-instruments for diagnostics purposes is allowed when the analytical performance specifications are fulfilled [14, 41]. External quality control of HbA1c- or blood glucose- measurements have not been implemented in comparable studies performed in pharmacies [24, 31, 32], even though this is recommended in guidelines [42, 43].

Eighty-seven percent of the participants were in the low-risk population and under the cut-off for HbA1c-measurement and only 1% were diagnosed with diabetes. One reason for this low risk profile could be that many of our participants were under 45 years old, while risk of T2D increases most after this age [1, 44]. The Diabetes UK is validated from 40 years old [18] and the FINDRISC test is validated for those who are 45 years or older [14]. Thus, 45 years old should have been the age limit for participation in the study. Including a higher share of participants with a non-European background would most likely have increased the percentage of participants with high risk of diabetes, since people with a non-European background have a higher prevalence of undiagnosed diabetes, also those living in Norway [1]. The share of participants with a non-European background varies from none to eight in similar studies [24, 32, 45]. In light of this, our 10% indicates that including a pharmacy in an area with a relativity high number of customers with a non-European background as well as offering the risk test in English as well as Norwegian might have had some effect on inclusion. One third (8 of 24) of the participants with a non-European background took the risk assessment in English, and they had probably not been included if the risk assessment was only available in Norwegian. We considered offering the risk assessment as well as the flyers and posters in languages such as Arabic and Somali in addition to English and Norwegian to increase the share of participants with a non-European background. However, we could not ensure that the pharmacist could convey relevant risk information to participants who could not communicate in English or Norwegian. Both different languages and other cultural differences have been found to be barriers for communication between health care providers and immigrants [46]. In addition, to consult with health care providers for chronic diseases screening and illness prevention is unfamiliar for a lot of immigrants, who are accustomed to only seek health care when they are ill [47]. Another limitation of the study was that we only have the age group of the participants, and not the exact age. Thus, the mean age of the participants could not be calculated. Consequently, that should be changed before a larger study is conducted.

The service was positively received by the participants, and 90% of responders would recommend it to others. However, due to the low response rate of the follow-up questionnaire (35%), the participants who answered may not be representative for all participants, and the results of the follow-up questionnaire must therefore be interpreted with caution. Possible explanations for this low response rate are that not all participants received the follow-up questionnaire or a reminder because of high workload and because not all the participants´ e-mail addresses were readable. Lack of time for the pharmacists was also an issue in a previous study [24]. In a larger study the pharmacist could either write the e-mail address themselves, or check the handwriting while the participant is present to ensure that the e-mail address is correct, and a reminder must be sent shortly after the deadline.

Conclusion

This study confirms that it is feasible to implement a diabetes risk assessment tool followed by HbA1c- measurement in community pharmacies in Norway. The service is feasible and acceptable to both pharmacies and responding participants. In a large-scale study the inclusion criteria should be increased to 45 years in accordance with the population the FINDRISC has been validated for, and to avoid inclusion of too many people at low risk of T2D. To ensure a successful implementation of the intervention in a large-scale study a customized training program must be provided, in addition to detailed instructions and follow-up by the project leader. Furthermore, the pharmacy must provide enough time for both testing and follow-up of the participants. The pharmacies must perform internal quality control and participate in external quality control program to ensure analytical quality of the HbA1c instrument.

Supporting information

Acknowledgments

We thank all the participants, the pharmacists and pharmacy who participated, as well as the Pharmacy chain Apotek 1, Noklus and Siemens that lent out DCA Vantage. This project was supported by The Norwegian Community Pharmacy Foundation and the University of Bergen.