Factors Associated with Follow-Up Adherence in Patients Seen at a Referral-Based Dermatology Clinic for the Homeless

To the Editor: Skin disease is prevalent in homeless populations and contributes to significant morbidity.1Montgomery A.E. Szymkowiak D. Marcus J. Howard P. Culhane D.P. Homelessness, unsheltered status, and risk factors for mortality: findings from the 100 ,000 homes campaign.Public Health Rep. 2016; 131: 765-772Crossref PubMed Scopus (41) Google Scholar, 2Stratigos A.J. Stern R. Gonzalez E. Johnson R.A. O'Connell J. Dover J.S. Prevalence of skin disease in a cohort of shelter-based homeless men.J Am Acad Dermatol. 1999; 41: 197-202Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar, 3Chen C.L. Fitzpatrick L. Kamel H. Who uses the emergency department for dermatologic care? A statewide analysis.J Am Acad Dermatol. 2014; 71: 308-313Abstract Full Text Full Text PDF PubMed Scopus (9) Google Scholar Data are lacking on the unique factors that influence successful continuity of dermatologic care for homeless persons. To identify demographic and clinical factors associated with loss to follow-up, we performed a retrospective analysis of patients seen between May 2009 and July 2017 at a referral-based dermatology clinic for homeless individuals (University of Utah institutional review board number 00096567). The clinic is located in Salt Lake City, Utah, a large urban center in a rural state, and it provides free medical care specifically for the homeless population. Dermatologists volunteer in twice-monthly clinics after referral from primary care providers. Medical data from patients 18 years and older were obtained from electronic health records (EHRs) and included demographics, diagnostic information, treatment, recommendations, follow-up information, and comorbidities. Demographics were compared by sex by using t tests for continuous variables and chi-square tests for nominal variables. Generalized estimated equations with exchangeable covariance structure were used to determine variables associated with follow-up. SAS, version 9.4 (Cary, NC), was used for all analyses. During the study period, 141 dermatology clinics were held during the study period serving a total of 507 individuals. The average age at first visit was 48.9 years, and 73% identified as white. There were more men (68.8%), and men were more likely to be older at first visit than women (P < .001). Most patients (86.2%) had a Charlson Comorbidity Index (CCI) of 1 or less; however, 66.9% had at least 1 mental health diagnosis; these rates were higher in women (P = .048) (Table I). Benign conditions (42.6%) and dermatitis/psoriasis (26.6%) were most commonly diagnosed. Infectious (P = .002) and premalignant/malignant (P = .03) skin diagnoses were more commonly diagnosed in men, whereas acne/rosacea (P = .009) was more commonly diagnosed in women.Table IDemographic and clinical characteristics of patients, overall and by sex, at a dermatology clinic for homeless individuals in Salt Lake City, UtahCharacteristicsTotalFemaleMaleP valuePatients, N507158349Age first visit, y, mean (SD)48.9 (10.5)46.0 (10.6)50.2 (10.1)<.001∗P < .05.Number of diagnoses per patient, mean (SD)2.0 (1.5).8 (1.2)2.0 (1.6).13Number of clinic visits per patient, mean (SD)1.2 (0.6)1.2 (0.6)1.2 (0.7).61Race, n (%).74 White370 (73.0)111 (70.3)259 (74.2) Hispanic or Latino/a68 (13.4)25 (15.8)43 (12.3) Black/African American37 (7.3)12 (7.6)25 (7.2) Other32 (6.3)10 (6.3)22 (6.3)Charlson Comorbidity Index, n (%).39 0306 (60.4)87 (55.1)219 (62.8) 1131 (25.8)45 (28.5)86 (24.6) 232 (6.3)14 (8.9)18 (5.2) 324 (4.7)8 (5.1)16 (4.6) 4+14 (2.8)4 (2.5)10 (2.9)Number of mental health diagnoses, n (%).048∗P < .05. 0168 (33.1)43 (27.2)125 (35.8) 1146 (28.8)47 (29.7)99 (28.4) 298 (19.3)27 (17.1)71 (20.3) 368 (13.4)30 (19.0)38 (10.9) 4+27 (5.3)11 (7.0)16 (4.6)Dermatologic diagnosis category, n (%) Benign216 (42.6)74 (46.8)142 (40.7).20 Dermatitis/psoriasis135 (26.6)37 (23.4)98 (28.1).27 Infection110 (21.7)21 (13.3)89 (25.5).002∗P < .05. Premalignant/malignant104 (20.5)23 (14.6)81 (23.2).03∗P < .05. Acne/rosacea38 (7.5)19 (12.0)19 (5.4).009∗P < .05. Other143 (28.2)46 (29.1)97 (27.8).76SD, Standard deviation.∗ P < .05. Open table in a new tab SD, Standard deviation. The average time to first follow-up was 1 year. Of the 246 patients for which follow-up was recommended, 49.6% followed up as recommended. Patient demographics associated with increased follow-up adherence included older age (P = .03), male sex (P = .001), more skin diagnoses (P = .03), premalignant/malignant diagnoses (P = .03), in-clinic procedure (P = .001), shorter recommended follow-up intervals (P = .001), and fewer mental health diagnoses (P = .02) (Table II). Specifically, those diagnosed with personality disorders were associated with follow-up adherence (P = .003), whereas those diagnosed with posttraumatic stress disorder were associated with decreased follow-up adherence (P = .01). These findings can aid clinicians in altering recommendations or identifying patients who may be at risk for potential nonadherence.Table IIComparison of variables between patients who did and did not follow up among patients for whom follow-up was recommended (n = 246)VariablePatient followed up?P valueNo (n = 96)Yes (n = 150)Age, y, mean (SD)47.6 (11.0)50.3 (9.2).03∗P < .05.Number of diagnoses per visit, mean (SD)1.8 (1.4)2.5 (2.1).03∗P < .05.Sex, n (%) Male48 (57.1)104 (78.8).001∗P < .05. Female36 (42.9)28 (21.2)Race, n (%).17 White64 (76.2)105 (79.5) Hispanic or Latino/a12 (14.3)15 (11.4) Black/African American2 (2.4)9 (6.8) Other6 (7.1)3 (2.3)Charlson Comorbidity Index, n (%).76 050 (59.5)78 (59.1) 121 (25.0)36 (27.3) 25 (6.0)10 (7.6) 35 (6.0)4 (3.0) 4+3 (3.6)4 (3.0)Number of mental health diagnoses, n (%).02∗P < .05. 027 (32.1)55 (41.7) 118 (21.4)44 (33.3) 2+39 (46.4)33 (25.0)Dermatology diagnosis category, n (%) Benign31 (32.3)62 (41.3).15 Dermatitis/psoriasis29 (30.2)40 (26.7).55 Infection18 (18.8)37 (24.7).28 Premalignant/malignant28 (29.2)65 (43.3).03∗P < .05. Acne/rosacea13 (13.5)19 (12.7).84 Other26 (27.1)42 (28.0).88In-clinic procedure, n (%).001∗P < .05. Yes34 (35.4)85 (56.7) No62 (64.6)65 (43.3)Prescribed medication, n (%).30 Yes57 (59.4)79 (52.7) No39 (40.6)71 (47.3)Recommended follow-up time interval, mo, n (%).001∗P < .05. ≤130 (33.0)80 (55.2) >1 to 645 (49.5)53 (36.6) >616 (17.6)12 (8.3)SD, Standard deviation.∗ P < .05. Open table in a new tab SD, Standard deviation. Given the transience of homeless populations, commitments to follow-up care is understandably difficult. Although our patient population is largely physically healthy as measured by the CCI, the mental health burden of homeless US adults is 46%, which is greater than the national average of 18.5% for the general US population.4National Institutes of HealthMental illness.https://www.nimh.nih.gov/health/statistics/mental-illness.shtmlDate: 2019Date accessed: April 22, 2019Google Scholar,5US Department of Housing and Urban DevelopmentThe 2010 annual homeless assessment report to Congress.https://www.hudexchange.info/resources/documents/2010HomelessAssessmentReport.pdfDate: 2010Date accessed: April 22, 2019Google Scholar The correlations between mental health diagnoses and likelihood of following up points to the compounded difficulty of patient adherence when grappling with psychiatric disease. Although this study has several limitations, including limited sample size, EHR accuracy, and generalizability to the US homeless population, our results aim to better inform dermatology clinicians on the management of this unique and vulnerable population. Future studies should characterize barriers to dermatologic care for homeless persons to develop effective interventions and treatment strategies.