In the article “Assessing family history of chronic disease in primary care,”1 which appeared in the January issue of Canadian Family Physician, errors were inadvertently included. The corrections appear below.
The proportion of patients self-reporting a family history (FH) of colorectal cancer (CRC) was reported incorrectly. The results portion of the abstract should have read as follows:
Among patients with positive FH, the following proportions of patients had that FH recorded in the EMR [electronic medical record] compared with the questionnaire: diabetes, 24% in the EMR versus 36% on the questionnaire, κ = 0.466; coronary artery disease, 35% in the EMR versus 22% on the questionnaire, κ = 0.225; breast cancer, 21% in the EMR versus 22% on the questionnaire, κ = 0.241; and CRC, 12% in the EMR versus 26% on the questionnaire, κ = 0.510.
This also affected the “Comparison of FH collection methods” section. Additionally, McNemar tests should have been used to compare EMR and self-report data; this affected the P value for an FH of breast cancer (BC) This section should have read as follows:
Slightly more than 20% of women were identified as having an FH of BC by either EMR (21%) or self-report (22%, P = .74). The proportion of patients with an FH of CRC was 12% in the EMR and 26% by self-report (P < .001).
The “Analysis” section should have read as follows:
Two-sided Fisher exact tests were used to examine differences between elevated and average risk patients (based on FH) in appropriate screening variables. Cohen κ statistics were used to determine concordance of the data sources. The McNemar test was used to compare ascertainment of FH information from the EMR compared with self-report.
Some numbers were inadvertently transposed in Table 3. The correct version of the table is reproduced here.
Table 3.
Concordance between FH from EMR data and self-reported FH
| CONDITION | FH FROM EMR DATA | SELF-REPORTED FH | κ STATISTIC* (95% CI) | |
|---|---|---|---|---|
|
| ||||
| NO | YES | |||
| Diabetes (n = 775) | No | 459 | 134 | 0.466 (0.40–0.53) |
| Yes | 41 | 141 | ||
| Coronary artery disease (n = 775) | No | 433 | 73 | 0.225 (0.15–0.30) |
| Yes | 175 | 94 | ||
| Breast cancer (women only, n =555) | No | 365 | 73 | 0.241 (0.15–0.34) |
| Yes | 69 | 48 | ||
| Colorectal cancer (n = 775) | No | 571 | 114 | 0.510 (0.44–0.58) |
| Yes | 5 | 85 | ||
EMR—electronic medical record, FH—family history.
Strength of agreement30: < 0 = less than chance; 0.01–0.20 = slight agreement; 0.21–0.40 = fair agreement; 0.41–0.60 = moderate agreement; 0.61–0.80 = substantial agreement; and 0.81–0.99 = almost perfect agreement.
Two-sided Fisher exact tests should have been used to examine differences between elevated- and average-risk patients. This affected the P values reported in Table 4. The correct version of the table is reproduced here.
Table 4.
Risk level based solely on FH and screening from self-report health questionnaire or EMR data: Familial risk level is taken from self-reported data; screening information is taken from either self-reported data or EMR data.
| CONDITION | AVERAGE RISK BASED ON FH, N/N (%) | ELEVATED RISK BASED ON FH, N/N (%) | FISHER EXACT P VALUE |
|---|---|---|---|
| Diabetes (n = 775) | |||
| • Ever had screening | 378/500 (76) | 221/275 (80) | .15 |
| • Up-to-date screening | 342/378 (91) | 144/221 (65) | < .001 |
| Breast cancer (n = 555) | |||
| • Ever had screening | 381/434 (88) | 107/121 (88) | > .99 |
| • Up-to-date screening | 341/381 (90) | 75/107 (70) | < .001 |
| CRC (n = 775) | |||
| • Ever had screening* | 416/669 (62) | 85/106 (80) | < .001 |
| • Had correct type of screening† | 349/416 (84) | 84/85 (99) | < .001 |
| • Up-to-date screening | 288/349 (83) | 74/84 (88) | .25 |
CRC—colorectal cancer, EMR—electronic medical record, FH—family history.
For CRC screening, average risk includes average and low risk; elevated risk includes mildly increased risk and moderate to high risk.
Correct CRC screening for moderate to high risk includes colonoscopy only.
The authors apologize for the errors and any confusion they might have caused.
Des erreurs se sont glissées dans l’article intitulé « Assessing family history of chronic disease in primary care »1, publié dans le numéro du Médecin de famille canadien de janvier. Les corrections apparaissent ci-après :
La proportion de patients qui ont signalé eux-mêmes avoir des antécédents familiaux (AF) de cancer colorectal (CCR) a été incorrectement rapportée. Dans le résumé, la section des résultats aurait dû se lire comme suit :
Parmi les patients qui avaient des AF positifs, on notait les proportions suivantes pour ceux dont les AF était consignés dans le dossier médical électronique (DME) ou dans le questionnaire : diabète, 24 % dans le DME contre 36 % dans le questionnaire, κ = 0,466; maladie coronarienne, 35 % dans le DME contre 22 % dans le questionnaire, κ = 0,225; cancer du sein, 21 % dans le DME contre 22 % dans le questionnaire, κ = 0,241; et CCR, 12 % dans le DME contre 26 % dans le questionnaire, κ = 0,510.
D’autres corrections ont été apportées dans le texte intégral de l’article en anglais. Les auteurs présentent leurs excuses pour ces erreurs et pour toute confusion qu’elles pourraient avoir causée.
Reference
- 1.Carroll JC, Campbell-Scherer D, Permaul JA, Myers J, Manca DP, Meaney C, et al. Assessing family history of chronic disease in primary care. Prevalence, documentation, and appropriate screening. Can Fam Physician. 2017;63:e58–67. Available from: www.cfp.ca/content/cfp/63/1/e58.full.pdf. Accessed 2017 Sep 6. [PMC free article] [PubMed] [Google Scholar]