This was an observational cross-sectional study. The detailed patient recruitment and data collection methods, the study process with patients with diabetes and nurse professionals at the health center, and the questionnaires’ contents are described in Navigator’s validation study protocol [22]. The data collection was accomplished between October 2018 and September 2019.

Adult (≥18 years of age) patients with diabetes scheduled for a nurse’s annual control appointment were recruited for the study, and the patients volunteering to participate provided informed consent. Patients with a disability preventing informed consent for participation (eg, Alzheimer disease, mental disability) were excluded.

The sample size and power calculation were based on results of a 36-item WHODAS 2.0 validation study of patients with chronic conditions in Europe. The calculation used a power of 80% and statistical significance of P=.05. The result of the calculation was a total sample size of 300 patients. The details of the sample size calculation are presented in our Navigator study protocol [22].

Patients’ characteristics and their distribution into the different groups by gender and age groups are described with frequencies and proportions. The Kruskal-Wallis test was used to compare the patients’ median age in the different groups [31,32]. The concordance and mismatch between the nurse’s evaluation of Navigator’s result and actual result are also described with proportions.

The Pearson chi-square test was used with categorized variables to compare the self-acting and other patient groups’ characteristics, smoking, chronic conditions and medication, and self-rated health. The distribution of scale variables (age distribution, years since the diabetes diagnosis, diabetes care values, WHODAS 2.0, EQ-5D-5L VAS, WBQ-12, and Navigator’s question “Do you know how to use electronic services?” on a VAS) were explored with the Kolmogorov-Smirnov test. For normally distributed variables, the independent samples t test was used to compare variable means, whereas the Mann-Whitney U test was used to compare medians if the variable was not normally distributed.

We describe frequencies of missing values that were excluded from the analyses. We analyzed the data with SPSS versions 25 and 26 (IBM Corp).

Tampere University Hospital’s Ethics Committee approved this study’s ethical aspects in October 2018 (ETL R18070). Data collection at Valkeakoski Health Center was approved by the head physician in September 2018. All patients and nurses participating in the study were provided with an informed consent form detailing the study aims and process, their voluntary participation, and ability to end participation without any consequences. We anonymized all data; thus, the data were examined without identifying information. We did not provide any compensation to the study participants.

Altogether, 538 patients were invited to the study, and 304 (56.5%) participated. Most participants (272/304, 89.5%) returned the study questionnaires. Nurses completed Navigator’s questionnaires with all 304 patients during an appointment. The majority of patients (232/304, 76.3%) were 60 years to 79 years old, and their gender distribution was balanced (Table 1).

Navigator segregated patients into the 4 groups, as follows: 259 into the self-acting group, 34 into the cooperating group, 6 into the community group, and 5 into the network group (Table 2).

In the self-acting group, the majority of patients (206/259, 79.6%) were 60 years to 79 years old, and their gender distribution was balanced. Patients were mostly married or in a relationship (160/236, 67.8%), and their school education was comprehensive (82/235, 34.9%) or vocational (93/235, 39.6%). The majority (199/236, 84.3%) were retired. Self-acting patients’ median age did not significantly differ from other patients’ median age; however, the proportion of patients ≥80 years old was smaller in the self-acting patient group (Multimedia Appendix 1, Table 3).

The self-acting patients had lived with a diabetes diagnosis for less than 10 years, and the result differed from the cooperation and network groups. The difference was statistically significant (P=.003; Table 4).

The self-acting patients’ HbA_1c values differed from those of the cooperation and network groups, and the difference was statistically significant (P<.001), as was the difference in LDL cholesterol results between groups (P=.04). The differences in other diabetes care values (UACR, systolic and diastolic blood pressures, BMI) and smoking status between groups were not statistically significant (Table 4).

Hypertension was the most prevalent comorbidity, and hyperlipidemia was the second most prevalent comorbidity in both the self-acting patient group and combined cooperation and network patient groups. Arthrosis, arthritis, or joint pain was the third comorbidity in both groups. Statistically significant differences between groups appeared in the prevalences of ischemic cardiac disease (P=.001), depression or anxiety (P=.03), asthma or chronic obstructive pulmonary disease (COPD; P<.001), and long-term pain (P<.001; Table 5).

In the self-acting patient group, hypertension, hyperlipidemia, joint ailment (arthrosis, arthritis, or joint pain), and long-term pain were also the most prevalent comorbidities assessed separately with each chronic condition (Multimedia Appendix 2).

Metformin was the most common diabetes medication in both the self-acting patient group and the combined cooperation and network patient groups, and there was no difference in the use of metformin between the groups. In the self-acting patient group, a DPP-4 inhibitor was the second most common medication (53/236, 22.5%), while in the combined cooperation and network patient group, the second most common medication was insulin or a biosimilar medication (16/31, 52%). This difference was statistically significant (P<.001; Multimedia Appendix 3).

Statistically significant differences were present in the amount of diabetes medication. In the self-acting group, 10.6% (25/235) of the patients did not have any medication for diabetes, 54% (127/235) of the patients used 1 drug, and 18.3% (43/235) of the patients used 2 drugs. In the combined cooperation and network patient group, all patients used diabetes medication, 29% (9/31) of the patients used 1 drug, and 52% (16/31) used 2 drugs (P<.001; Multimedia Appendix 3).

ACE inhibitors or AT2 receptor blockers were the most common antihypertensive drugs in both groups. Statistically significant differences between the groups were found in the usage of β-blocker, nitroglycerin, or digitalis medications, which were analyzed as 1 group (P<.001); pain medication (P=.03); psychopharmacological drugs (P=.001); and medication for pulmonary diseases (P=.004). The self-acting patients used less of these medications than the patients in the cooperation and network groups (Multimedia Appendix 4).

Most self-acting patients (130/232, 56%) were satisfied with their health, while 12.9% (30/232) were unsatisfied. In all other groups combined, 53% (18/34) were unsatisfied, and a minority (2/34, 6%) were satisfied. The difference between the self-acting patient group and all other patient groups was statistically significant (Pearson chi-square test, P<.001).

The responses of the self-acting patient group and all other patient groups to the WHODAS 2.0 differed by 12 points, which is a 20% difference in the medians, a statistically significant difference (P<.001; Table 6).

The difference in medians between the self-acting group and the other groups to the EQ-5D-5L index score values was 0.20 points and to the EQ VAS was 23.82 points, and both differences were statistically significant (P<.001). The clinical significance of the differences could be computationally determined with the minimally important difference (MID=0.5 x pooled SD) [33]. In our population, the computational MIDs were 0.066 for the EQ-5D-5L index value and 8.48 for the EQ VAS (Table 6).

In addition, statistically significant differences between the groups were detected in the WBQ-12 questionnaire’s dimensions of negative (P<.001), positive (P<.001), and general well-being (P=.004). To assess the clinical significance of the difference in the means for general well-being, we calculated the MID (MID=2.14; Table 6).

We explored Navigator’s criterion validity by comparing the nurses’ evaluation of the patient group with the actual result that Navigator proposed. Almost all of Navigator’s results (248/256, 96.9%) for the self-acting patient group were in concordance with the nurses’ evaluations. Two-thirds (67%) of the evaluations were in concordance with the actual Navigator result in the cooperation (24/36) and network (2/3) groups, while the concordance was less than one-third (2/7, 29%) in the community group. One-half (13/26, 50%) of the mismatches were in the evaluation of the patient’s medical state, and 31% (8/26) of the mismatches were in the evaluation of the patient’s functional ability in everyday life. In 19% (5/26) of the mismatching cases, both of these 2 dimensions were evaluated differently (Multimedia Appendix 5).

Patients responded to Navigator’s question “Do you know how to use electronic services?” on a VAS of 1 (yes) to 10 (no). The self-acting patient group’s median of 1 (Q1=1 and Q3=3; n=259) differed from the other groups’ median of 2 (Q1=1 and Q3=9; n=45). When analyzed with the independent samples Mann-Whitney U test, the difference was statistically significant (P=.001).

Navigator assigned most patients to the self-acting group. The patients in the self-acting group had comorbidities and medication in addition to diabetes but fewer of both than the patients in the cooperation and network groups. The self-acting patients differed from other patient groups in disability, health-related quality of life, most dimensions of well-being, and self-rated health. The proportion of patients over 80 years old was 3 times higher in the other patient groups than in the self-acting group, though the groups’ median ages did not differ significantly. Almost all patients considered their skills at using electronic services to be good. Additionally, the agreement between the professionals’ evaluations of the patient group with the actual result that Navigator proposed (criterion validity) was very high for the self-acting patients.

Most self-acting patients were 60 years to 79 years old, and they had had diabetes for less than 10 years. The prevalence of diabetes doubles in the Finnish population at the age of 60 years to 69 years compared with younger age groups [34], and the prevalence peaks in high-income countries in those aged 75 years to 79 years [24], which may reflect our findings regarding the age distribution. The shorter duration of diabetes may impact the self-acting patients’ diabetes care, as they had less diabetes medication, and oral agents were emphasized when compared with the situation of patients in the combined cooperation and network groups. Diabetes care seemed to be well-balanced, as most of the self-acting patients reached all the general targets of diabetes care (HbA_1c<53 mmol/mol, LDL cholesterol<2.6-2.8 nmol/L, blood pressure<140/80) [35,36]. The differences in HbA_1c and LDL cholesterol values between the 2 groups may be associated with the different target values in care. The HbA_1c target for older patients rises as their disability reduces [37]. Additionally, the LDL cholesterol target is based on the duration of diabetes (less or more than 10 years), patient’s age, and related comorbidities [35-37].

The self-acting patient group’s most common comorbidities were hypertension, hyperlipidemia, and arthrosis, which is in line with previous findings in multimorbidity research [38,39]. Ischemic cardiac disease, depression or anxiety, and asthma or COPD were more prevalent in the cooperation and network group, which may be associated with the increase in the incidence and prevalence of diabetic complications and multimorbidity in patients of an older age and with a longer history of diabetes [40,41]. These results conform to our findings of the self-acting patients’ medication, as the patients’ smaller number of comorbidities result in a reduced need for medication compared with the patients in the cooperation and network group.

The self-acting patients’ lower number of comorbidities and medication may in turn reflect their better self-rated health, functional ability, health-related quality of life, and well-being, when compared with all other groups.

We noticed a 12-point difference (20% in medians) when assessing disability with the 12-item WHODAS 2.0. A clinically significant difference of 9 points was observed in a Finnish population with chronic musculoskeletal pain [33], as was a change of 5% to 10% in scoring in hospitalized patients [42,43]. Though previous studies were conducted in different settings, we may consider our greater differences clinically significant in favor of the self-acting patients’ better functional ability. This result is also in line with the findings on multimorbidity, defined as the presence of 2 or more chronic conditions, and its association with reduced ability. Functional decline is greater with a higher number of conditions as in the combined cooperation and network patient group, especially when depressive symptoms are included [38,41,44,45].

The differences between the self-acting and all other groups in health-related quality of life measured with the EQ-5D-5L and EQ VAS exceeded the MID values by almost 3 times. When assessing patients with diabetes and chronic pulmonary diseases, the MID in the EQ-5D-5L index score values may vary from 0.03 to 0.069 [46,47] and in the EQ VAS may vary from 0.5 to 9.7 [47,48]. Thus, we may consider our results as clinically significant in favor of the self-acting patients’ better health-related quality of life. Previous findings have shown that patients’ diabetes-related clinical conditions, poor glycemic control, injectable medication, and polypharmacy negatively affect health-related quality of life, which in our results may appear as the other patient groups’ lower values in health-related quality of life [27,49-54].

Health-related well-being was measured with the WBQ-12, and the self-acting patients’ better general well-being score compared with those of the other groups may be clinically significant, as the differences in means between groups again exceeds the MID [33]. In our population, the self-acting patients’ general well-being was a little lower compared with that of patients in other cultures and populations with newly diagnosed type 2 diabetes and those with a greater than 12-year history of the disease [55-57].

Additionally, the results on self-rated health conform with the previous differences between groups in functional ability, health-related quality of life, and general well-being. Aging and multimorbidity have been associated with reduced self-rated health as well, and they presumably impact our findings [26,45,58].

All patients, not only the self-acting patients, responded that their ability to use electronic services was good. Therefore, advising and empowering all capable patients about utilizing digital equipment and applications in their self-care could improve patient care and care results, release health care resources for patients needing services at face-to-face appointments, and reduce health care costs [1,3-6,8-10]. In addition to the capability, the patients’ willingness, concerns, and fears regarding digital services should be noted, especially with older or otherwise vulnerable patients [12,13,16,17]. Digital alternatives in health services should supplement and support the patient’s individual care instead of complicating and thus harming it.

The concordance of the nurse’s evaluation of the patient’s Navigator result and the actual result was very high for the self-acting patients. The information in the patients’ EHRs and some nurses’ and patients’ relational continuity of care may facilitate patient knowledge and impact the result positively [59,60].

The small number of patients in the other groups complicates the interpretation of the overall criterion validity result. However, the high mismatch percentages in the 3 small groups may indicate that, other than self-acting patients, patients are not easy to identify by a professional. This emphasizes the importance of paying attention to the patients’ views of their values and coping in their life. Previous results concerning Navigator’s user experiences indicated that Navigator’s questions eased raising issues in conversation and helped the professionals to extensively understand the patient’s general care. The questions assisted the patients in understanding their situation better and to see their situation from new perspectives [61]. Thus, the use of Navigator probably impacts the group results, explaining the mismatches. In general, Navigator might be especially useful with patients who do not have notes in EHRs and new patients with newly diagnosed morbidities.

The number of patients to evaluate the patients’ characteristics was sufficient only in the self-acting group. Although the community, cooperation, and networks groups were merged, they still formed a small sample of other than self-acting patients. Significant differences between groups were noted, but individual extreme responses may be emphasized especially in small groups and thus bias the group result. Additionally, some missing values in the questionnaire responses may bias the result. Therefore, our study results of the group differences in medical condition and disability, health-related quality of life, well-being, and self-rated health are intended for generating new hypotheses and need to be confirmed in future studies.

The vast proportion of patients being assigned to the self-acting group may be due to the capability of these patients to attend follow-up appointments at a health center, while older patients may be placed in home health care instead of visiting a health center. Our distribution result may also derive from the patients’ voluntary participation in the study based on their willingness and attitudes toward medical research [62,63] and older and sicker patients’ denial of informed consent, which leads to refusal to participate in a study [64]. This refusal may have reduced the number of patients in the presumed older and sicker patient groups.

This study was performed only at one health center, the study participants were volunteers, and the distribution of this population with diabetes into the self-acting patient group was emphasized. Therefore, this limits the generalizability of our results. Additionally, the differences between groups were noticed particularly between all patients; therefore, at the individual level, a group result for a single patient cannot be determined deductively on the grounds of this study.

In assessing the criterion validity, only 1 nurse evaluated the patient’s incoming Navigator result as in real life. The concordance result may have been different if another professional had conducted the evaluation. The forthcoming reliability study may hone the results, as the Navigator questionnaire was later repeated by the same nurse and also by a physician simultaneously.

Our results support the hypotheses that the self-acting patients were younger at the group level; had a simpler medical condition; and had better self-rated health, functional ability, health-related quality of life, and well-being. As far as self-acting patients form the largest group of patients with diabetes, their care pathway could be designed to account for their multimorbidity and to take a holistic approach in care integration, emphasizing the approach of preventing diabetic complications and disabilities. Digital services as complementary options for supporting self-care and an alternative to health center visits may be integrated into all care pathways and offered to all patients willing to use them. Efficiently targeting needs-based health services to various patient groups could reduce the worldwide burden and related costs of aging and multimorbid populations in health care. Navigator may add value besides databased patient segmentation methods, as the patients’ own views of their capabilities in everyday life are particularly widely considered with Navigator.