Skip to main content
Version: v0.1.10

Tutorial: process a lab feed

In this tutorial we build a small program that takes a lab's batch file, reads every result in it, lists the abnormal ones, and answers each message with an acknowledgment. Each step is a complete program you can paste into a file and run with Node.js 22 or later after npm install @cosyte/hl7. Every message is synthetic.

Step 1: split the batch file​

A lab sends its results as one file: a batch header (BHS), one message per patient, and a trailer (BTS) that declares how many messages it holds. splitBatch returns each message already parsed.

import { splitBatch } from "@cosyte/hl7";

const feed = [
  "BHS|^~\\&|LAB|MAIN|EHR|REF|20260419150000",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143000||ORU^R01^ORU_R01|LAB-0001|P|2.5",
  "PID|1||MRN10001^^^HOSP^MR||Doe^Alice||19750620|F",
  "OBR|1|ORD-1|FLR-1|BMP^Basic Metabolic Panel^L",
  "OBX|1|NM|GLU^Glucose^LN||182|mg/dL|70-99|H|||F",
  "OBX|2|NM|NA^Sodium^LN||139|mmol/L|135-145|N|||F",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143500||ORU^R01^ORU_R01|LAB-0002|P|2.5",
  "PID|1||MRN10002^^^HOSP^MR||Roe^Bob||19620311|M",
  "OBR|1|ORD-2|FLR-2|CBC^Complete Blood Count^L",
  "OBX|1|NM|HGB^Hemoglobin^LN||10.1|g/dL|13.5-17.5|L|||F",
  "BTS|2",
].join("\r");

const batch = splitBatch(feed);

batch.messages.length; // => 2
batch.warnings; // => []

batch.warnings is empty because the trailer's count, 2, matches the messages found.

Step 2: read each patient's results​

Each entry is either { ok: true, message } or a failure. For each parsed message, read the patient's record number and every result with observations().

import { splitBatch } from "@cosyte/hl7";

const feed = [
  "BHS|^~\\&|LAB|MAIN|EHR|REF|20260419150000",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143000||ORU^R01^ORU_R01|LAB-0001|P|2.5",
  "PID|1||MRN10001^^^HOSP^MR||Doe^Alice||19750620|F",
  "OBR|1|ORD-1|FLR-1|BMP^Basic Metabolic Panel^L",
  "OBX|1|NM|GLU^Glucose^LN||182|mg/dL|70-99|H|||F",
  "OBX|2|NM|NA^Sodium^LN||139|mmol/L|135-145|N|||F",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143500||ORU^R01^ORU_R01|LAB-0002|P|2.5",
  "PID|1||MRN10002^^^HOSP^MR||Roe^Bob||19620311|M",
  "OBR|1|ORD-2|FLR-2|CBC^Complete Blood Count^L",
  "OBX|1|NM|HGB^Hemoglobin^LN||10.1|g/dL|13.5-17.5|L|||F",
  "BTS|2",
].join("\r");

const results: string[] = [];
for (const entry of splitBatch(feed).messages) {
  if (!entry.ok) continue;
  const mrn = entry.message.patient?.mrn ?? "(no MRN)";
  for (const obs of entry.message.observations()) {
    results.push(`${mrn} ${obs.identifier.identifier} ${String(obs.value)}`);
  }
}

results; // => ["MRN10001 GLU 182", "MRN10001 NA 139", "MRN10002 HGB 10.1"]

The glucose value is the number 182, not the string "182": OBX-2 says the value is numeric (NM), so the helper returns a number.

Step 3: list the abnormal results​

OBX-8 carries the lab's abnormal flag: H for high, L for low, N for normal. Keep the results whose flag is anything other than N or empty.

import { splitBatch } from "@cosyte/hl7";

const feed = [
  "BHS|^~\\&|LAB|MAIN|EHR|REF|20260419150000",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143000||ORU^R01^ORU_R01|LAB-0001|P|2.5",
  "PID|1||MRN10001^^^HOSP^MR||Doe^Alice||19750620|F",
  "OBR|1|ORD-1|FLR-1|BMP^Basic Metabolic Panel^L",
  "OBX|1|NM|GLU^Glucose^LN||182|mg/dL|70-99|H|||F",
  "OBX|2|NM|NA^Sodium^LN||139|mmol/L|135-145|N|||F",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143500||ORU^R01^ORU_R01|LAB-0002|P|2.5",
  "PID|1||MRN10002^^^HOSP^MR||Roe^Bob||19620311|M",
  "OBR|1|ORD-2|FLR-2|CBC^Complete Blood Count^L",
  "OBX|1|NM|HGB^Hemoglobin^LN||10.1|g/dL|13.5-17.5|L|||F",
  "BTS|2",
].join("\r");

const abnormal: string[] = [];
for (const entry of splitBatch(feed).messages) {
  if (!entry.ok) continue;
  const mrn = entry.message.patient?.mrn ?? "(no MRN)";
  for (const obs of entry.message.observations()) {
    const flag = obs.abnormalFlags ?? "";
    if (flag !== "" && flag !== "N") {
      const units = obs.units?.identifier ?? "";
      abnormal.push(`${mrn} ${obs.identifier.text ?? ""} ${String(obs.value)} ${units} (${flag})`);
    }
  }
}

abnormal; // => ["MRN10001 Glucose 182 mg/dL (H)", "MRN10002 Hemoglobin 10.1 g/dL (L)"]

Step 4: answer every message​

The lab expects one ACK per message. buildAck echoes the message's control id so the lab can match the answer. Accept a message that parsed and carries its results; answer AE (application error) when it is missing the OBR an ORU^R01 requires. A message that could not be parsed at all has no control id to echo, so we set it aside for a person to look at.

import { buildAck, interpretAck, splitBatch } from "@cosyte/hl7";

const feed = [
  "BHS|^~\\&|LAB|MAIN|EHR|REF|20260419150000",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143000||ORU^R01^ORU_R01|LAB-0001|P|2.5",
  "PID|1||MRN10001^^^HOSP^MR||Doe^Alice||19750620|F",
  "OBR|1|ORD-1|FLR-1|BMP^Basic Metabolic Panel^L",
  "OBX|1|NM|GLU^Glucose^LN||182|mg/dL|70-99|H|||F",
  "MSH|^~\\&|LAB|MAIN|EHR|REF|20260419143500||ORU^R01^ORU_R01|LAB-0002|P|2.5",
  "PID|1||MRN10002^^^HOSP^MR||Roe^Bob||19620311|M",
  "MSH|^",
  "BTS|3",
].join("\r");

const answers: string[] = [];
const setAside: string[] = [];

for (const entry of splitBatch(feed).messages) {
  if (!entry.ok) {
    setAside.push(entry.error.code);
    continue;
  }
  const missing = entry.message.structure.missingSegments;
  const ack =
    missing.length === 0
      ? buildAck(entry.message, { code: "AA" })
      : buildAck(entry.message, { code: "AE", error: { conditionCode: "100", severity: "E" } });
  const answer = interpretAck(ack);
  answers.push(`${answer.controlId ?? ""} ${answer.code ?? ""}`);
}

answers; // => ["LAB-0001 AA", "LAB-0002 AE"]
setAside; // => ["MSH_TOO_SHORT"]

The second message arrived without its OBR, so msg.structure.missingSegments names it and we answer AE with condition code 100 (segment sequence error, HL7 Table 0357). The third is not a message at all, and the batch kept going past it.

What we built, and where to go next​

The program splits a batch, reads typed results, flags the abnormal ones, and answers every message it can. In production the feed arrives over MLLP; @cosyte/mllp receives it and sends the ACKs back.