Evaluation of a Direct-to-Patient Telehealth Service in Germany (docdirekt) Based on Routine Data

Video consultations have become a promising additional means of healthcare provision in recent years, a trend accelerated by the COVID-19 pandemic (1, 2). However, no evaluation of patients’ uptake of these direct telemedicine services has yet been performed for acute care in Germany.

In April 2018 the Association of Statutory Health Insurance Physicians in Baden-Württemberg (KVBW) initiated the first exclusively remote care service for statutorily insured persons (docdirekt). This 2-year project enabled telemedical consultation and treatment with no prior in-person physician contact (3). Patients could contact a physician between 09.00 and 19.00 from Monday to Friday via chat, telephone call, or video call (for more details see www.docdirekt.de).

The aim of this analysis was to assess whether this form of telemedicine can reduce attendances in doctors’ offices and emergency rooms.

Methods

We analyzed anonymized routine data from the docdirekt software comprising all telemedical consultations that occurred from April 16, 2018, to April 16, 2020. The outcome quality of the telemedical consultation was extracted from the physicians’ written case notes by means of a two-stage process. The first step involved automated identification of cases that contained certain words documented by the physician (e.g., “hospital,” “emergency”) indicating the need for further treatment in a hospital. In the second step, these cases were scrutinized and divided into three categories: primary resolution by telemedicine (including the recommendation to visit a doctor’s office in the event of deterioration or persistence of the underlying conditions, or for further medical evaluation and treatment) (I); the recommendation to go to the hospital urgently if the symptoms should deteriorate or persist (II); and the recommendation for immediate admission to the hospital (including emergency services) (III). All cases that could not be assigned to categories II or III in step 1 were defined as category I. The differences between cases with the various consultation results were investigated by means of subgroup analysis.

Furthermore, we analyzed data from a patient survey conducted by the KVBW for internal quality control purposes 2 weeks after the telemedical consultation.

Results

A total of 3090 cases were treated telemedically via docdirekt (table 1). Most users were between the ages of 18 and 39 years, followed by the 40 to 64 age cohort. The telephone was the medium most often used for the teleconsultation (73.3%), followed by chat (26.4%) and video call (18.2%). The majority of patients lived in urban areas (>80%). Most cases could be resolved by means of telemedicine (88.2%). In 5% of cases the patients were recommended to present urgently to a hospital should the disease worsen or persist, while immediate hospital attendance was recommended in 6.7%. If consultation was by video, the odds of further treatment being classified as more urgent were lower by a factor of 0.65. Two weeks after telemedical consultation, 428 patients (response rate: 13.9%) took part in a survey (table 2). In more than 66% of these cases, the reason for the consultation was definitively resolved from the patient’s point of view. About three fourths of the participants stated that they would have used other healthcare services if telemedical consultation had not been available. Almost all patients (97.7%) would recommend docdirekt to others.

Table 1. Results of subgroup analysis for patients with different consultation outcomes on the basis of routine data from the docdirekt software.

	All	Subgroup (I)	Subgroup (II)	Subgroup (III)	Odds ratio*2	95% confidence interval (odds ratio)*2
Variable	n = 3090	Recommendation for
		Resolved telemedically	Hospital in event of deterioration	Immediate hospital admission
		n = 2726 (88.2%)	n = 156 (5.0%)	n = 208 (6.7%)
	n/observed (%)	Cases in the subgroups (%)
Sociodemographics
Female	1573 (50.9)	53.8	49.4	50.8	1.05	[0.85; 1.31]
Male	1498 (48.5)	45.7	50.0	48.6	0.95	[0.77; 1.19]
0–17 years	598/3088 (19.4)	12.5	33.3	19.1	1.11	[0.85; 1.44]
18–64 years	2014/3088 (65.2)	51.4	53.2	67.0	0.54	[0.43; 0.67]
65 years or older	476/3088 (15.4)	36.1	13.5	14.0	2.34	[1.81; 3.02]
Rural	372/1929 (19.3)	22.3	18.4	19.0	1.13	[0.83; 1.52]
Urban	1557/1929 (80.7)	77.7	81.6	81.0	0.89	[0.66; 1.20]
Contact medium
Chat*1	815 (26.4)	11.1	21.2	27.8	0.46	[0.35;0.62]
Telephone*1	2266 (73.3)	96.6	94.2	70.4	9.19	[5.53;15.27]
Video*1	561 (18.2)	10.6	16.7	18.8	0.65	[0.47;0.89]

*¹ Since in some cases different media were used within one consultation, the sums of the cases and the percentages exceed the total number of cases (n = 3090) and 100%, respectively.

*² The odds ratios and their confidence intervals were extracted from separately estimated proportional odds models for ordinal logistic regressions, with the urgency of further treatment (consultation outcome I, II, or III) serving as the (ordinally scaled) dependent variable. Interpretation: If, for instance, the consultation was by telephone, the odds were higher by a factor of 9.19 that further treatment was considered less urgent.

Table 2. Results of the patient survey for internal quality control purposes 2 weeks after telemedical consultation.

Variable	n = 428 (%)
Sociodemographics
Female	229 (53.5)
Male	199 (46.5)
Age (mean, in years)	37.1
Regular primary care physician (yes)	402 (98.3)
Was it possible to complete the consultation via docdirekt?
Yes	283 (66.1)
Would you recommend docdirekt to others?
Yes	418 (97.7)
What would you have done if telemedical ‧consultation had not been available?
Gone to an emergency room	202 (49.9)
Gone to a doctor’s office	89 (22.0)
Waited	83 (20.5)
Looked on the internet	13 (3.2)
Called the emergency services	9 (2.2)
Other	9 (2.2)

Discussion

The docdirekt project was designed to reduce avoidable attendances in doctors’ offices and emergency rooms in the German state of Baden-Württemberg by creating an infrastructure for treatments carried out exclusively via communication technology. The results of this study indicate that the infrastructure created may be capable of achieving this goal. However, the service was taken up mainly by patients in urban areas.

In most cases no referral to other healthcare facilities was needed. Moreover, the results of the post-consultation survey suggest that many patients would have attended other healthcare services such as doctors’ offices or emergency rooms/hospitals if the docdirekt service had not been available. Comparable studies have had similar findings (1, 4).

Furthermore, our analysis showed that most patients did not use video (less than 20%), revealing a clear preference for telephone calls. Although telephone consultations can be effective for some medical problems (4), findings reported in the literature suggest that there is still potential to improve the communication skills needed for telephone consulting (5). Physicians providing telemedical treatment should be aware that patients may prefer to communicate by telephone. The peculiar features of this form of care provision should be taken into account (2).

In this study, the outcome quality of the consultation was derived from the physicians’ case notes and from the post-consultation patient survey. Routine data on patients’ subsequent treatments or health resource utilization were not taken into account.