APPENDIX
Diabetes case finding methodology
We applied a tiered approach to search for diabetes and diabetes medication-related keywords within the four text fields in the medical examination forms using regular expressions. Regular expressions allowed us to account for differences in spaces in between words, capital and small letters, word boundaries, and line breaks. First, we changed abbreviations such as N.I.D.D.M. and D.M. to NIDDM and DM, respectively and deleted certain words such as “DMPA” (depot medroxyprogesterone) and “cytoproteronidm” in order to avoid capturing the “DM” components of those phrases and misclassifying them as diabetes cases. Then we searched the text field for diabetes phrases that were clearly describing diabetes cases (e.g., known DM, known patient of diabetes, a case of NIDDM, has had diabetes) (Appendix Table 1a).
If no match was found in the text field, the next step in the algorithm was to search for diabetes-specific medications in the categories of sulfonlyureas, biguanides (metformin), insulin, and general oral hypoglycemic agent keywords (Appendix Table 1b). Phrases describing insulin resistance and hyperinsulinemia were accounted for and not considered a match for a diabetes keyword. Polycystic ovary syndrome (PCOS) is often treated with metformin so we searched for phrases such as “polycystic”, “ovary”, and “PCO”. If a PCOS keyword was found in conjunction with only a medication in the biguanide class, the phrase was not considered a match for diabetes.
If no match was found in the first two steps, the text field underwent a final stage of processing. First, the field was split into phrases based on the presence of certain punctuation marks such as a period, semicolon, colon, hyphen, question mark, forward/backslash, and line break. The text within each of those phrases was parsed, searching for simple diabetes keywords (e.g., diabet, mellitus, IDDM, DMII) (Appendix Table 1c). Then, we searched for negations in the same phrases (e.g., no diabetes, no history, rule out, negative for diabetes) to make sure that we were not misclassifying refugees as having diabetes (Appendix Table 1c).
We applied this tiered approach to three of the four text fields on the DOS forms. For the one medication-specific text field, we only searched for medication and diabetes-related keywords and did not apply negation terms. We aggregated across the four text fields to create a composite diabetes variable and considered the refugee as having diabetes if a match occurred in any of the four fields. We manually checked results at each step and iteratively updated the search terms to ensure that our algorithms captured and negated appropriate keywords. Finally, as a quality control check, we compared our diabetes definition with self-reported diabetes from the medical history section of the DS-3026. We were in agreement in 4,344 diabetes cases and we found an additional 791 cases using our text parsing algorithms. However, 843 refugees self-reported diabetes that we did not capture. Upon further investigation, 632 (75.0%) had nothing written in the text fields which did not allow them to be considered as cases by our definition. The other 211 had comments noted in at least one of the four text fields. The 632 self-reported diabetes cases were included as cases.
In some instances, text fields in the database contained “see scanned document”, presumably when the writing was illegible and could not be manually entered. In 187 (<0.1%) of 249,037 records, all four of the text fields that we evaluated for our diabetes case definition contained this statement, thus not allowing them to be considered as diabetes cases. These 187 records were excluded from our analysis.
Appendix Table 1. Regular expressions code used to capture cases of diabetes among adult US-bound refugees, Electronic Disease Notification System 2009-2014
a. First step in capturing cases of diabetes*"(?<!not a )known DM|(?<!not a )known diabetic|(?<!no )known diabetes|(?<!not a )known patient of diabetes|(?<!not )a known case of DM|Diabetes Mellitus, known|known case o(f|d) diabetes|A case on NIDDM|A case of NIDDM|A case of NID DM|Has had diabetes mellitus|Diabetes Mellitus since|diabeties mellitus since|DMsince|Type II diabetic since|diabetessince|diabeticpatient|NIDDMdiagnosed|diabetes mellitus diet controlled|diabetes mellitus for|diabetes mellitus x|NIDDMfor|diabetes mellitus, recent diagnosis|Diabetes mellitus not very well controlled|Diabetes mellitus on no meds|Diabetes mellitus type II since|Diabetes mellitus, for|diabectomynellitus type II, since|\\ndiabet|^diabet|\\n.\\)\\s*Diabet|\\nNIDDM|^NIDDM|\\nDM(?!PA)|^DM(?!PA)|\\nHypertension, diabetes|^Hypertension, diabetes|\\nHypertension and diabetes|^Hypertension and diabetes|\\nHistory of hypertension and diabetes|^History of hypertension and diabetes|\\nH/O Hypertension \\(.*\\) & Diabetes|^H/O hospitalization for HTN and DM|^Hypertension; Diabetes|^Hypertension; Hypercholesterolemia; Diabetes|\\nHistory of DM|^History of DM|\\nHistory of diabetes mellitus|^History of diabetes mellitus|\\nHx of diabetes mellitus|^Hx of diabetes mellitus|-History of diabetes|-diabetes|-\\s*diabetes|-NIDDM|no diabetic compliations|no diabetic complication|nodaibeticcomplication|no app(ea|a)r(e|a)nt diabetic complication|no DM complications|nno DM comlications|no diabetic compliocations|nodiabetccomplications|noapperant sign of daibeticcomplication|no diabetes complication|no evidence of DM complication|R/O diabetic neuropathy|no diabetic or hypertensive changes|no diabetic retinopathy|with diabetic neuropathy|class b other\\s*:\\s*diabet|class b others\\s*:\\s*diabet|class b other\\s*:\\s*DM|class b others\\s*:\\s*DM|class b other\\s*:\\s*NIDDM|class b others\\s*:\\s*NIDDM|class b others:\\s*NID\\s*DM|class b type 2 diabetes|class b other\\s*:\\s*A case of DM|class b other\\s*:\\s*A case of Diabet|class b other\\s*:\\s*A case of NIDDM|\\nHx, of Gestational DM|\\nwell controlled Diabetes|^\\s*DIABETES|\\nUnderlying disease of D(M|iabet)"
b. Second step in capturing diabetes-specific medications*
sulfonylureas = "Glipizid|Glucotrol|Gliclazide|Glicazide|Diamicron|Glibenclamid|Glibendamid|Glybenclamid|Glibenclemid|Glibendmid|Gliberclamid|Glibincalmid|Gilberclamid|Glibencalamid|Glibonclamid|Glibeclamid|Glyburid|Diabeta|Micronase|Glynase|Glimepirid|Amaryl|Acetohexamide|Chlorpropamide|Diabinese|Diabenese|Tolbutamide|Orinase|Diamet\\b|Glimepride|Glitazone|Gliclazid|Diamicron|Diabeton"
biguanides = "Metform|Metaformin|Metoformin|Melformin|Metforum|Meformin|Fortamet|Glucophage|Glycophage|Glucophag|Glucuphage|Glumetza|Riomet"
insulin = "insulin|\\bNPH\\b|Humulin|Humulog|Novolin|Novolog|Lantus|Solostar|mixtard|Novoram"
hypoglycemic agent = "on\\s*oral\\s*hypoglycemic|on\\s*oral\\s*hypoglacaemic|on\\s*hypoglymeic\\s*agent"
c. Final step in searching for diabetes keywords and negation terms*
diabetes keywords= grep("diabet|diabletic|dibbetic|diabit|diabtes|Idabetes|Dibetes|Dibees|mellitus|melitus|millitus|militus|IDDM|\\bNIDDN\\b|\\bDM(?!PA)(?!irtrate)|NIDDDM\\b|NDDM\\b|IIDM\\b|INDDM\\b|NIDMM|\\bNIDM\\b| INDDA|NIDDD|NIDDN|\\bDDMI\\b|DMII\\b|\\bIDM|\\bGDM\\b"
negations = "no\\s*diabet|no\\s*diabit|no\\s*IDDM\\b|no\\s*NIDDN\\b|no\\s*DM\\b|no\\s*NIDDDM\\b|no\\s*NDDM\\b|no\\s*IIDM\\b|no\\s*INDDM\\b|no\\s*GDM\\b|no\\s*history|no\\s*hx|no\\s*h/o|no\\s*prior\\s*history|no\\s*prior\\s*hx|no\\s*prior\\s*h/o|no medical hsitory|nodiagno|\\br/o\\b|rule\\s*out|ruledout|under evaluation for Diabet|notdiabetic|never gave history of|no past history of|not have past history|tested for diabetes|no signs/symptoms of D(M|iabet)|no S/S of D(M|iabet)|no signs/symptoms of hypertension or diabetes|no signs/symptoms and diabetes|no signs or symptoms of DM|no signs of diabetes|no signs of hypertension or diabet|no sign and sympotm of DM|no signs and symptomp of DM|no sign of hypertension or diabet|no signs of hypertension and diabetes|no signs and symptoms of diabetes|non-d|non- d|nond|non -d|non - d|no sign of diabetes|no evidence of hypertension or diabetes|no evidence of diabetes|no evidence of HTN , HF or DM|no evidence of DM|No evidence of dibetesmellitus|notfonfirmed to be diabetic|not hypertensive or diabetic|denieshx of diabetes|deniesDM|denies any history of hypertension/diabetes|no hypertension, diabetes|denied|Ho history of|no previous history|no documented history|he hasn't got|nohistry|not associated with|denied the history or symptoms|negative for diabetes|negative for sugar|\\bordm|\\bordiab|noany|do not give history|no other associated|not confirmed to be diabetic|no clinical evidence of diabet|DOES NOT HAVE DIABETES|never had diabet|no GDM detected"
* (?<!”text” ) = no match if “text” comes before the phrase; (?!”text”) = no match if “text” comes after the phrase; \\b = word boundary; \\n = start of a line; ^ =start of a text string; \\s* = white space (0 to many spaces); .* = wildcard; special characters that require \\ = . ( ) \
Appendix Table 2. Age group by body mass index category interaction with diabetes among US-bound refugees, Electronic Disease Notification System Jan 2009–Aug 2014*
Age group in years / Body Mass Index categoryNormal / Overweight / Obese
OR (95% CI)† / OR (95% CI) / OR (95% CI)
Near East
18-44 / ref / ref / ref
45-64 / 15.2 (12.1, 19.2) / 9.5 (8.1, 11.1) / 7.4 (6.4, 8.5)
65-74 / 40.3 (31.2, 52.2) / 22.1 (18.3, 26.6) / 16.1 (13.6, 19.1)
75+ / 39.7 (28.8, 54.8) / 17.8 (13.6, 23.4) / 12.2 (9.3, 15.9)
Western Hemisphere
18-44 / ref / ref / ref
45-64 / 16.2 (7.3, 36.3) / 5.6 (3.3, 9.4) / 4.1 (2.5, 6.8)
65-74 / 48.1 (20.3, 114.2) / 16.1 (9.0, 28.9) / 11.9 (6.5, 22.0)
75+ / 19.8 (5.7, 68.4) / 24.8 (11.9, 51.6) / 7.3 (2.3, 22.6)
South and Central Asia
18-44 / ref / ref / ref
45-64 / 12.1 (9.3, 15.7) / 5.6 (4.2, 7.6) / 7.9 (4.5, 14.1)
65-74 / 16.8 (12.2, 23.1) / 10.7 (7.0, 16.2) / 7.4 (2.4, 23.1)
75+ / 15.5 (10.2, 23.5) / 16.8 (9.7, 29.4) / 15.9 (4.8, 52.4)
* Models control for sex age group, BMI category, history of tuberculosis, and year of refugee US-arrival
† Odds ratio (95% confidence interval)